• Aucun résultat trouvé

Operationalizing meta-approaches in IS : the case of institutional framworks

N/A
N/A
Protected

Academic year: 2021

Partager "Operationalizing meta-approaches in IS : the case of institutional framworks"

Copied!
27
0
0

Texte intégral

(1)

CAHIER DE RECHERCHE : 2008-07 E5

Operationalizing meta-approaches in IS : the case of

insti-tutional framworks.

CARTON Sabine, DE VAUJANY François-Xavier,

ROMEYER Cécile

Unité Mixte de Recherche CNRS / Université Pierre Mendès France Grenoble 2 150 rue de la Chimie – BP 47 – 38040 GRENOBLE cedex 9

(2)

Operationalizing meta-approaches in IS : the case of

in-stitutional frameworks

Sabine CARTON *

François-Xavier DE VAUJANY** Cécile ROMEYER***

*Maître de conférences - CERAG UMR 5820 - IAE Grenoble Sabine.Carton@iae-grenoble.fr

**Professeur- CERAG UMR 5820- IAE Grenoble devaujany@iae-grenoble.fr

***Maître de conférences - Chercheur Préactis- ISEAG- Université Jean Monnet Cecile.romeyer@univ-st-etienne.fr

(3)

ρρρρ Abstract

IS research's growing interest in institutional and neo-institutional approaches is confirmed by several literature reviews. Nonetheless, few works have studied the instrumentation of institutional frameworks, the concrete way these very meta-theoretical perspectives have been implemented up till now. This research begins with an overview of institutional and neo-institutional research, along with the more intermediate frameworks which have been extrapolated from them, before an analysis of a set of empirical articles or communications (from 1999 to 2006). Data have been treated by means of a thematic, a lexicometric and a cluster analysis (based on a canonical analysis). Thematic and lexicometric analysis show the domination of cross-sectional research (in particular for communications), and some interesting specificities in the process and level of analysis. Most research thus assumes some transnational and universal mechanisms of legitimation of technology (whatever the theoretical framework implemented). More specifically, the cluster analysis results in the elaboration of three groups of empirical research, with for all of them different instrumenta-tion strategies.

Keywords:

(4)

Introduction: about the breakthrough of institutional approaches in IS

The world of information systems has been experiencing numerous fads, fashions and imi-tations. In the late 90's, firms massively invested in e-business tools in order to not 'miss the train'. A couple of years later, others chose to launch costly Enterprise Application Integra-tion (EAI) or Enterprise Resource Planning (ERP) projects in order to follow ineluctable tendencies in their industry. More recently, some companies have done their best to imple-ment so-called "best practices" concerning IS project manageimple-ment and process integration, contributing to the diffusion and banalization of best practices. Best practices have thus be-come common practice, without any differentiating power. In these phenomenon, the com-munication of competitors, commercial actions of software producers or hardware manu-facturers, and those of professional networks or unions, play a major role. This is precisely the goal of institutional approaches in IS, to make sense of these external constraints and theories and their influence on the design, implementation and use of Information Tech-nologies in organizations (Orlikowski and Barley, 2001). The idea is to show that most firms include these constraints in a quest for legitimacy.

Emerging in the 90's, institutional approaches appealed to a wider audience in IS (Or-likowski and Barley, 2001; Mignerat and Rivard, 2005). Beyond a purely rational and in-ternal vision of IS-related practices, they shed light on the factors and process which influ-ence the management of information technologies in organizations.

Several meta-analysis have already been carried out by researchers of the IS field in order to understand the structuration of this literature (Mignerat and Rivard, 2005). They show that the bulk of institutional works in IS rely on Scott (2001) or DiMaggio and Powell (1991), in particular their notions of "institutional pillars", "isomorphism", "constraints" and more rarely, "organizational fields" (Mignerat and Rivard, 2005). The problem of the instrumentation of these (meta) theoretical frameworks and the operationalization of their major constructs has rarely been treated. In continuation to Walsham (1993) about structu-ration theory, a meta-theoretical approach can be defined either as a framework with a high level of abstraction and generality (thus implying a more intermediate framework to make sense of empirical data), or as a framework developed by its author without any methodo-logical concerns. Institutional approaches, as structuration theory, can be viewed as meta-theoretical frameworks.

For this research, we would like to answer the following research question: what are the strategies implemented in IS to instrument/operationalize institutional approaches?

The first part of the paper is devoted to an introduction to general institutionalist and neo-institutionalist frameworks, along with neo-institutionalist applications or intermediated frame-works of the IS field. Then, our research method based on lexicometric, thematic and facto-rial analysis is detailed. It relies on the study of all articles indexed on ABI and paper con-ferences (from ICIS) hosted on the Aisel portal for the 1999-2006 period. It is followed by a presentation and discussion of our main results.

1.The main institutionalist or neo-institutionalist frameworks implemented in IS re-search

It is hard to present institutional streams (along with their related intermediate frameworks) without understanding key notions such as "institution", or more generally, the seminal

(5)

in-stitutional frameworks. We will then illuminate the difficulty to instrument these theoretical lenses.

1.1. Institutional and neo-institutional approaches: an overview

Institutional and neo-institutional works do not really correspond to a clear-cut and ho-mogenous field of research. In order to clearly introduce this network of theory, we will first define the notion of institution, and we will then describe the main streams of this lit-erature.

What is an institution? As a pre-requisite, Searle (2005 : 2-3) first suggests understanding the notion of "institutional facts", and more precisely that "In some intuitively natural sense, the fact that I am an American citizen, the fact that the piece of paper in my hand is a 20 dollar bill, and the fact that I own stock in AT&T, are all institutional facts. They are institutional facts in the sense that they can only exist given certain human institutions. Such facts differ from the fact, for example, that at sea level I weigh 160 pounds, or that the Earth is 93 million miles from the sun, or that hydrogen atoms have one electron". Our life is thus full of institutional facts, which only make sense in view of the institution they bear. Thus, for Searle, "An institution is any collectively accepted system of rules (procedures, practices) that enables us to create institutional facts. These rules typically have the form of X counts as Y in C, where an object, person, or state of affairs X is assigned a special status, the Y status, such that the new status enables the person or object to perform func-tions that it could not perform solely by virtue of its physical structure, but requires as a necessary condition the assignment of the status. The creation of an institutional fact is, thus, the collective assignment of a status function." (Searle, 2005: 21). An institution thus implies both a system of rules and a framework for equivalencies, which make it possible to locate and compare objects and people to each other.

In a similar manner, Scott (2001) defines an institution as a "social structure" which gives organizations and individuals lines of action and orientations, by controlling and constrain-ing them. Insistconstrain-ing also on constrainconstrain-ing powers of institutions, Barley and Tolbert (1997: 94) state that an institution: "represent(s) constraints on the options that individuals and col-lectives are likely to exercise, albeit constraints that are open to modification over time". But what are the main institutional and neo-institutional approaches? Among the works originating this field of research, it is first possible to mention Commons (1931, 1950) and Veblen (1898). For Commons, an institution is a framework of laws or natural rights, a "collective action in control, liberation and expansion of individual action" (1931, p 648). From his point of view, the smallest unit of economic activity is what he calls a "transac-tion" (a well-known notion which will become the basis of Williamson’s theory). It inter-venes "between the labour of the classic economists and the pleasures of the hedonic economists, simply because it is society that controls access to the forces of nature, and transactions are, not the "exchange of commodities," but the alienation and acquisition, be-tween individuals, of the rights of property and liberty created by society, which must therefore be negotiated between the parties concerned before labour can produce, or con-sumers can consume, or commodities be physically exchanged…". For Commons, action is at the heart of the production and reproduction of order and "operating rules", which refers "to and emphasizes the reproduction of working rules etc (the structural aspects of the transaction) through such activity, whilst 'conflict' (and its resolution) is the main dynamic for change or transformation. » (Lawson, 1994: 195). Eventually, if institutions have a real constraining power in Commons writings, they do not determine the nature of actions. They boil down to giving collective action its underpinnings.

(6)

Likewise, Thorsten Veblen (1898) has insisted on the notion of habits of thought structured by institutions (which ultimately live inside us and structure our way of thinking). In the meantime, he also rejects any form of determinism ("As Thorsten Veblen (1898) insisted, these habits of thought provide both the conceptual frameworks of meaning and enquiry, and the systems of values that ground our inspirations ", Hodgson, 1999: 28).

Among the other seminal authors, Marx, Berger, Luckman, Mead, Cooley and Selznick are extremely significant (Scott, 2004). Their contributions to the definition and conceptuali-zation of institutions and social structures have been used by many neo-institutionalists, both in the fields of economics, and sociology and organization theory1. Indeed, all of these fields have appropriated more or less explicitly the institutionalists’ heritage by applying it to a new research object: organization.

In the field of economics, Williamson (1975) has used the notion of "transaction" elabo-rated by Commons to extend some aspects of Coase's research (1937) about the nature of the firm2. The project was to overcome the classic neo-institutionalist axiomatic so as to explicitly include institutions in economics analysis (and the different institutional shapes firms can endorse). One of transaction cost economists key research questions (also known as "the new institutional economics) was the comparison between the cost of market use and those of organizations (Gomez, 1996; North, 1990).

Obviously, institutional legacy (that of the so-called old institutionalism) is rarely claimed by transaction cost economists3.

In the field of sociology, the relationship and continuity between old institutionalism and neo-institutionalism is implicit but real (Scott, 2004 and Hodgson, 2004) and the key con-cepts of Commons and Veblen have been a bit forgotten4. The idea is to represent organiza-tions as linked together and constructed by their environment (Rojot, 2005). According to neo-institutionalists (in sociology), organizations belong to common fields ("organizational fields") where populations of organizations experience some common constraints. Beyond a quest for efficiency and effectiveness, organizations also hanker for their legitimacy with regard to customers, suppliers, government and society as a whole. This leads them to con-form to some isomorphic pressures (DiMaggio and Powell, 1983, 1991). Isomorphism means "the constraining process that forces one unit in a population to resemble other units that face the same set of environmental conditions (DiMaggio and Powell, 1991: 66). Ac-cording to DiMaggio and Powell5, these pressures can be threefold: coercive (i.e. linked to the pressure exerted by third-party regulations), mimetic (in a situation of uncertainty, or-ganizations tend to reproduce the legitimate behaviour of other oror-ganizations) or normative (i.e. the result of a normative process) At the micro-social level, organizations can then adopt five kinds of strategies to cope with these isomorphic constraints: acquiescence, compromise, manipulation, avoidance and defiance (Oliver, 1991).

After a period during which neo-institutionalist emphasised external and constraining social structures, some promoters of the stream now tend to defend a conceptualization of social structures which is quite close to Giddens' (1984)' view. Scott (2004) thus insists on the vir-tuality of social structures, only instantiated through agency.

1 The institutionalist field is also applied in politics science (see: Moe, 1984), but no re-used in management science

2

Quoted by Wiliamson in his 1975's book.

3

Veblen and Commons are rarely mentioned by Coase and Williamson, who nonetheless suggested a return to the semi-nal ideas developed by institutiosemi-nalists.

4

The major book including DiMaggio and Powell’s 1991 chapter "does not even include the name of Commons in its index" (Rojot, 2005: 419). But it mentions Selznick (1969) several times.

5

(7)

Today, neo-institutionalism in sociology can be divided into several streams, all structured around distinct leaders. (Scott, 2004) On the one side, DiMaggio and Powell. On the other, Scott, Meyer and Rowan. The former have instrumented the concept of institution by means of the notion of social network. The latter have relied on symbolic models. Another way to shed light on the various sub-streams is the level of analysis: macro or micro (Scott, 2004).

Ultimately, institutions, these very specific social structures examined by institutionalism, have been the subject of numerous analyses in the fields of economics, sociology, organiza-tion theory, or human resource management. In the field of Informaorganiza-tion systems research, numerous studies have also targeted institutional dynamics.

1.2. Major Institutional and Neo-institutional Frameworks Implemented in IS

Let us now come back to Searle's analysis (2005) about institutions. According to econo-mists, the life of institutions and collective action implies: a collective intent (i), an assign-ment of functions (ii) and a functional status (iii).

There is no collective action without a project. Furthermore no collective action exists without a role assigned to everybody and a status related to these functions. This is the sense of the notion of functions which is particularly relevant for IS. According to Searle (2005: 7), "Again, human beings have a capacity that they share with some, though this time with not very many, other species of animals, the capacity to impose functions on ob-jects where the object does not have the function, so to speak, intrinsically but only in vir-tue of the assignment of function. "The study of an object of IS (ERP, knowledge manage-ment systems, groupware technologies, or even an organizational IS on the whole) from an institutional perspective would thus aim at understanding the functional assignment real-ized by a broad, inter-organizational community. Gradually, IS objects make sense in the course of their institutionalization within equivalencies systems, and sometimes become real institutions (this is probably the case today with ERP technologies). More generally, institutional works in IS try to explain "how institutions influence the design, use, and con-sequences of technologies, either within or across organizations" (Orlikowski et Barley, 2001: 153).

Since 1989, Kling and Iacono have emphasised the importance of the representation of IS as a set of institutions in order to explain the reasons for IS project failures. These authors put forward several arguments justifying the relevance of institutional approaches: the same computer tool can be integrated and appropriated in different manners according to the or-ganization context. One of the key success factors for an IS project would be its "utilisabil-ity", more than its technical capability to treat information. More over, well appropriated and stable IS, deeply embedded into organizational social structures, would be much more complex to replace than those less developed with fewer users. Kling and Iacono cannot but underline the irony of the situation: "we invoke institutional analysis to understand com-puterization in action" (Kling and Iacono, 1989: 23).

Likewise, Tornatzky and Fleischer (1990) insist on the societal dimension of technology. For these authors, technology is seen as a tool following a function on the basis of knowl-edge and principles (Tornatzky and Fleischer, 1990). Thus, they state: "it is critical to re-member that technology, old or new, is at its heart a cognitive construct. Without knowl-edge, there is no technology. While our emerging definition of technology as (a) function-performing tool based on knowledge and principles has a surface plausibility, there are shortcomings to this definition. There are many artefacts that perform functions or

(8)

innova-tive practices in our culture that do not usually get considered as technologies. (…). " (Tor-natzky and Fleischer, 1990: 13).

But if some researchers in the field of IS (like Kling and Iacono) explicitly mention found-ing authors like Selznik, Meyer or Rowan, others use more intermediate models which have been designed specifically for illuminating IS problems.6 Such is the case of the organizing vision model offered by Ramiller and Swanson (1997, 2002), or King et al’s (1994) institu-tional approach to IS.

It is possible to introduce the two aforementioned middle range theories by means of the

following table:

6

Some IS articles based their theoretical foundations on a mix of institutional approaches with structuration theory (. Bennett Thatcher, Brower, and Mason 2006; Barrett, Sahay et Walsham, 2001,Hirt et Limayem 2003) or with Actor Network theory (Backhouse, Hsu et Silva, 2006) or both (Chae and Poole, 2005).

MIDDLE RANGE THEORIES

Authors King et al 1994 Swanson and Ramiller 1997 Institutional sources and main

features of the theory

Synthesis of perspectives on the relationship between innovations and institutions.

Proposition of a theoretical frame-work including the possibility of institutional interventions in the field of innovation

Di Maggio and Powell (1991), Scott (1991), Selznick (1957)

Emphasized the key role of some external stakeholders like consultants, IT practitio-ners, or software manufactur-ers in the adoption of a tech-nology in an organization. Selznick (1957), DiMaggio and Powell (1991), Scott (2000)

Role of institutions Institutions are forces of regulation or influence of IT

Institutions influence the adoption of a technology at a micro level, that of organiza-tions, which in turn feed the evolution of institutions.

Object of research Analysis of the influences on inno-vation with regard to IT

Study of adoption processes of IT innovations in firms

Authors using the framework in their own research

- Damsgaard J., Lyytinen K. (2001) (about the adoption of EDI technol-ogy);

- Montealegre R. (1999) (on the adoption of Internet in developing countries);

- Damsgaard J., Scheepers R. (1999) (on the process of intranet implementation);

- Silva L., Figueroa E. (2002), (on the macro-economics development of IT).

Wang, (2001); Firth, (2001); Wang and Swanson (2007); Carton, de Vaujany and Romeyer, (2007); Bureau, (2006)

(9)

• Table 1: Main features of two major middle range theories about the role of in-stitutions in IS

1.3. About the Difficulties to Instrument Institutional Frameworks in IS

In the continuation to institutional and neo-institutional approaches, middle range theories of IS give few landmarks to researchers interested in their instrumentation. The organizing vision model clearly epitomizes this problem (Carton et al., 2003, 2007). If Swanson and Ramiller (1997) have worked out a subtle framework, likely to illuminate various ICT-related practices, they remain rather silent on their modalities of operationalization. This problem can also be extended to other institutional IS models, where methodological as-pects are rarely treated7.

Among the rare works which have proposed a panorama of instrumentation strategies of institutional approaches, Mignerat and Rivard’s (2005) literature review can be mentioned. By means of a study of a sample of papers published in top-tier international journals, they distinguish causalist approaches from other more processual approaches. On the whole, they remarked the domination of cross-sectional research ("However, most studies re-viewed looked at IT phenomena, with an institutional lens, using cross sectional method-ologies", Mignerat and Rivard, 2005: 15). Their analysis also led them to notice that these approaches are rather "counter-intuitive and inappropriate to institutional analysis, where process research should be the most appropriate approach. " (see also Robey and Bou-dreau, 1999 on this criticism).

2. Research method

First, the way in which data have been collected will be detailed. Then, we will explain the coding and treatment we applied to these data.

2.1. Data Collection (of articles and conference papers)

Our first objective was to find articles and communications 'applying' institutional or neo-institutional (general or middle range) frameworks to IS objects. We relied on three data-bases for data collection: EBSCO and ABI (for articles) and Aisel portal (for communica-tions). On Aisel, we collected papers in the proceedings of ICIS (International Conference on Information Systems). The use of these ICIS documents represented a significant set of papers, and we believe that the choice of a national or continental conference (like AMCIS, ACIS or ECIS) would introduce some redundancies in the communication database.

By means of request on the body of the text8, we started by identifying a set of texts corre-sponding to our core target (institutional or neo-institutional papers with an empirical sec-tion)9. Papers then had to be the subject of a cross validation between the authors of this study. Numerous papers (more than half of them for the communication database) were re-moved.

2.2.Coding of Articles and Communications

7

Other meta-theoretical perspectives are also very hard to instrument, like structuration theory (Jones, 1999) or critical realism (de Vaujany, 2008). In both cases, the operationalization of several key constructs is very hard.

8

"Institutionalism"+"information system", "institution"+"information system", "institution"+"information technology", "institutionalism"+"technology", "institutional"+"technology". Query based on titles + abstract + keywords + full text.

9

(10)

The set of remaining papers and communications were then coded by means of a thematic coding as described by Miles and Huberman (2002). The thematic dictionary has been worked out collectively by using the following procedure (as described notably by Weber, 1990):

i) First detailed definition of a preliminary set of categories, in order to start with a first common representation for the coding.

ii) A first sample of papers (20) has been coded by each co-author, which has resulted in several collective discussions about the categories implemented. This test has notably been an opportunity to refine the definition of some categories, and also to add new categories. The test ended once researchers obtained a consensus on the thematic dictionary and the way it should be implemented. The preliminary test contained 8 articles. It was a landmark for the remaining coding process.

iii) Individual coding of the other articles and communications, by means of the prelimi-nary test and the precise definition of each category identified.

iv) Comparison between the 49 articles and 36 communications (85 articles on the whole) selected in order to check and confirm the relevance of the coding scheme implemented by the three researchers.

The categories included in the thematic dictionary are the following:

CODE (name of the variable) MODALITIES COMMENTS Methodological approaches

(ap-prométho)

Action-research, survey, case study, experiment, meta-analysis (i.e. analysis of a set of previous empiri-cal research), hybrid approach (combination of the previous ones). Level of analysis (them) Societal, inter-organizational,

intra-organizational.

Period of publication (Date) Year of the publication of the article or communication

Target (Géné/spé) Specific technical concept, Informa-tion technologies or InformaInforma-tion Systems on the whole

Nature of data collection (Datcoll) Punctual or longitudinal

Nature of data treatment (Nattrait) - First generation of quantitative techniques: simple or multiple lin-ear regression, descriptive tests, etc; - Second generation quantitative techniques: structural equation modelling (like LISREL or PLS), factorial analysis, dependency net-work diagrams, logistical regres-sion, game theory techniques… - Unstructured qualitative methods (no coding and thematic dictionary was mentioned);

- Structured qualitative methods (with a coding and treatment of the coding)

When the paper did not detail the way in which qualitative data analy-sis was achieved, we assumed that it

was an unstructured data analysis. We first wanted to create a hybrid category (corresponding to combi-nations of our four main methods).

But we did not really find hybrid papers. All clearly fitted into a

dominant research strategy.

(11)

num-pint) ber of empirical papers including interculturality into analysis. Sectorial approach

(Appcompsecto-rielle)

Intra-administration industry, intra-manufactory industry, intra-agrarian industry, intra-service industry, cross-sectorial

Theoretical framework instru-mented/implemented (cadtheo)

Mainly or partly institutional, Mainly or partly neo-institutional in the field of sociology, Mainly or partly institutional in the field of economics, mixt or with an implicit positioning.

When it was explicit in the paper, we followed the positioning sug-gested by the authors themselves.

Type of review/journal management, IS management or others.

• Table 2: Final thematic dictionary 2.3 Data Analysis

Three kinds of treatment have been implemented (see table 3 for a global presentation of the research method): a descriptive analysis of our thematic coding, a lexicometric analysis of the titles used for communications and articles, and a cluster analysis based on a canoni-cal analysis.

Firstly, the coding of the titles of articles and conference papers was the object of a descrip-tive data analysis (see appendix 1). The objecdescrip-tive was to count codes and their evolution over years.

Then, titles from the article and conference papers were treated by means of a lexicometric analysis. It is a widespread treatment in the French-speaking field of social sciences (Bardin, 1998). It consists of counting occurrences and co-occurrences (pivot analysis) of some keywords (except the so-called 'tool words'10 (see Bardin, 1998). Most frequent key-words are ranked, and some elements of context are also identified by means of pivot-analysis. Most software implementing lexicometric analysis also make it possible to lem-matize keywords (for instance "computer" and "computers" will be considered as the same word) or to identify repeated segments (such as "information systems". Two broad lexi-cometric analyses have been done: one on the whole database of titles (articles and confer-ence papers), without a distinction about the period of publication. Another on each period (based also on the whole database). The objective was to trace the evolution of keywords and their co-occurrence (see appendix 2). In order to realize the aforementioned treatments, we used a software called Sphinx Lexica TM.

Lastly, a cluster analysis was also implemented by means of a canonical analysis (see Valette-Florence 1998 for a justification of the tandem analyses). Canonical analysis (Overals) is a general form of factorial analysis which makes it possible to avoid some typical biases, such as a black-box effect which factorial analysis is typically blamed for (Tennenhaus, 1993). It enables also to perform cluster analysis on initial variables that are heterogeneous (nominal versus continuous scales). In order to achieve this kind of treat-ment, some preliminary modifications on the data are required: recoding of some modali-ties whose numbers of answers were too low, mergers of article and conference papers da-tabases (and thus n=85, 10 variables with 37 modalities) and elaboration of sets of variables before final analysis (see the following figure).

10

(12)

• Diagram 1: Preliminary works on initial data to perform canonical and cluster analyses

Some modalities were then recoded. For example in the variable “sectorial approach” see table 2 thematic dictionary, the following modalities were recoded in one modality: administration industry, manufactory industry, agrarian industry and service industry. The output in this case for this variable was only two modalities: intra-sectorial or cross-intra-sectorial. See appendix 4 to get an exhaustive view of the whole recoding process.

The merger of the conference papers base and the journals one was necessary to obtain a significant amount of articles in the population.

The constitution of sets of variables was done by means of several successive iterations which made it possible to shed light on three sets. These three sets were identified based on the meanings from the variables: some variables were used to define the method use, others to describe publications, and the last one enables to specify the research topic.

- Set 1: variables describing the method used in communications or papers (methodological approach, nature of data collection, nature of data treatment),

- Set 2: variables describing publications (period and type of publications),

- Set 3: variables describing the research topic (level of analysis, target, sectorial approach, topic, theoretical framework implemented, use or not of an international comparison .

(13)

Objective of the research Method Results Analysing the methodological

choices in order to treat institutional approaches

Distribution of the number of pa-pers in accordance with the catego-ries included in the thematic dic-tionary

Identifying main IS topics institu-tional approaches deal with

Analysis of the content of titles (of articles and conference papers)

Identifying group of articles using similar instrumentation strategies

-- Emphasis of dimensions structur-ing data

- Valorisation of group of papers, according to broad instrumenta-tion/operationalization of institu-tional frameworks

- Validation of the cluster analysis: confirmation of the discrimination of dimensions and validation of the classification index of modalities

• Table 3 : Presentation of the research method 3. Main results

First, we will detail the main results of our descriptive analysis, then those of our lexi-cometric treatment, before developing the results of the cluster analysis.

3.1.Descriptive analysis of the thematic coding

The analysis of the EBSCO and ABI databases resulted in the identification of 49 papers matching the criteria we choose.

Before detailing the results of this first treatment, a first interpretation related to the data itself struck us: the regular increase of empirical papers over the last nine years, in particu-lar for 2001 and 2006. 2006 alone represents 28,6% of the total number of publications. About the coding of articles, it seems that case-study based research largely outnumbers other kinds of studies (around 50%). Surveys represent 30% of our final database (but dominate in the A level publications, we found). Empirical approaches such as action re-search11 or experiments are rare.

For the level of analysis (inter-organizational, intra-organizational or societal), the distribu-tion of papers is the following: 40% for inter-organizadistribu-tional, 28,6% for intra-organizadistribu-tional and 30,6% for societal. These results are not really surprising, as most institutional and neo-institutional approaches invite researchers to look at social dynamics from a macro perspective (at the level of "organizational fields" or "populations of organizations"). The supremacy of inter-organizational analysis can also be explained (for a sociological ap-proach) by the idea of emphasising a certain organizational diversity so as to then make sense of a possible convergence ("isomorphism") in the field.

11

Totally absent from our sample.

Lexicometric analysis: number of occurencies, pivot analysis

Descriptive analysis

Cluster analysis (ward and Kmeans) enabled by a canonical analysis

ANOVA test and discrimi-nant analysis

(14)

Sixty three percent of the articles adopt a longitudinal perspective12 and 36,7% of the pa-pers use a structured qualitative method. More generally, qualitative treatments outnumber other treatments (they represent 61% of our sample). With regard to quantitative analysis, first generation analyses (24%) are more frequent than second generation treatments (14%). Interestingly, very few papers (only 16%) drawing on an institutional perspective use it to compare IS-related practices in various countries. Institutionalization process is thus as-sumed to rely on universal mechanisms. Conversely, half of the papers we found (51%) use cross-sectorial analysis. Industries are thus assumed to rely on specific institutional mecha-nisms (but are ultimately common to every country).

The main theoretical framework we found implemented in the empirical literature was the sociological stream of neo-institutionalism (DiMaggio and Powell). It was implemented either fully of partly as an element of a more general model. The so-called "new institu-tional economics" surprisingly accounts for only 6,1% of the papers we found. The main institutional papers drawing on empirical data were thus largely sociological.

Lastly, our descriptive treatment of the article database shows that 71% of the papers are published directly in IS journals (such as MISQ, ISR, ISJ…).

Concerning communications, we found some interesting differences between this database and that of articles. The analysis of ICIS papers for the years 1999 to 2006 helped us to se-lect 36 empirical papers instrumenting an institutional or neo-institutional framework (on the basis of a first of 742 communications).

The descriptive analysis of the results of the communication database (carried out with the same request and the same coding) result in slightly different conclusions (than those we draw from that of the article database). Would the review process favour qualitative ap-proaches over quantitative ones? this difference is confirmed when we compare the articles and the conference papers database. Indeed, surveys amount to 62% of the total of papers we found, much more than case study analyses (24%). This is quite low compared to the results of the article database (which represented more than half of the set of papers). Like the article database, other methodological approaches are rare and in some cases (ac-tion research), completely absent. Thus, quantitative treatment is the most widespread ap-proach (73% of our communication database). But the conference we used to build our sample (ICIS) may be a bit specific. For quantitative treatments, second generation ap-proaches are much more common here (48%) than those of the first generation (24%). Again, the supremacy of inter-organizational perspective is obvious (57%).

But the regularity we observed in the progression of empirical papers for articles is not true for ICIS, even if ICIS 2006 represents 19% of the total number of communications.

In agreement with Mignerat and Rivard’s (2005) study, cross-sectional approaches clearly outnumber (by 70%) longitudinal ones. But again, international comparisons are extremely rare (only 11%).

As for article database, inter-sectorial approaches are also very frequent (75%).

Lastly, neo-institutional approaches are more present than the others (73%), but this time with a clear advantage for the new institutionalism in economics (43%) compared to socio-logical frameworks (30%). But finally, article and communication databases share a lot of common points.

12

Two points are worth mentioning. Firstly, we included all levels of publications (not only top tier journals) as other studies which emphasized the presence of cross-sectional approaches. Then, the bulk of the surveys we found in our sample used a cross-sectional approach.

(15)

3.2.Discourse Analysis: the Lexicometric Treatment of Articles and Communications From a lexicometric perspective (without a distinction between years), several interesting keywords were very frequent in the titles: outsourcing, development, implementation, fash-ion, impact. Articles dealing with institutional dimensions would thus target precise aspects of IS development: the development and implementation phases. Major topics we found in the keywords were adoption, diffusion, trust and outsourcing of IS.

The lexicometric analysis we did year by year also shows that outsourcing (with keywords such as outsourcing, offshore, BPO) combined with words borrowed from the field of eco-nomics (like agency, auctions, governance, transaction) are more present for years 2004, 2005, 2006. From 1999 to 2003, problematics such as adoption, diffusion, and assimilation of IS dominate (with keywords like assimilation, implementation, implementing, adoption, diffusion).

Lastly, the number of keywords per year keeps increasing: (10 keywords in 1997 and 156 in 2006), which is linked both to the increase in the number of articles, but also to their di-versity. The number of keywords (except tool words) thus increased more quickly than that of the number of papers.

Thus, from our results, IT phenomena analyzed through institutional frameworks fall under several general themes: IT innovation, IT development and implementation, IT adoption and use, and IT outsourcing. Compared to the results from Mignerat; Rivard (2005), we found a new topic issue which is outsourcing.

3.3.Typology of Empirical Articles and Communications: cluster analysis via a Ca-nonical Analysis

The results introduced in this section show the highlighting of two main dimensions struc-turing our population by means of a canonical analysis. Based on the results found in the canonical analysis (coordinates of modalities along the two dimensions), a cluster analysis helped us to put forward three groups of articles and conference papers corresponding to three specific instrumentation strategies of institutional approaches. These three groups were then validated by means of an ANOVA and a discriminant analysis, before a final in-terpretation.

3.3.1. The dimensions put forward by the canonical analysis

The choice of the number of dimensions relevant for a cluster analysis was carried out by means of a scree test, identifying 2 dimensions (with an inflexion point after these two di-mensions). These two dimensions correspond to 64% of the variance explained (1,284/2 dimensions) see table 4.

Dimension Sum 1 2 Lo ss Set 1 ,354 ,424 ,778 Set 2 ,401 ,351 ,751 Set 3 ,252 ,367 ,619 Mea n ,335 ,381 ,716

(16)

Eigen-value ,665 ,619

Fit 1,284

• Table 4: The two dimensions identified by the canonical analysis 3.3.2. Identification and validation of our groups by the cluster analysis

A cluster analysis of the coordinate by means of Ward technique resulted in three groups (see dendrogram of appendix 3.2). The number of groups has then been confirmed by the Kmeans method, which made it possible to better visualize the modalities assigned to each group (see diagram 2).

Validation of the cluster analysis was carried out by means of two tests.

An ANOVA (test F) showing that the three groups were significatively different on the two dimensions (sig 0,000). Besides, a discriminant analysis of the coordinate validated the three group classifications, with a classification index of 97% (see appendix 3 to get the ta-bles).

Lastly, the table below (table 5) shows relative homogeneity in the distribution of modali-ties within these three groups.

Cluster 1 11,000

2 18,000

3 8,000

Valid 37

Missing 0

Table 5: Number of modalities in each cluster

3.3.3. Interpretation of the cluster analysis

The three groups identified by means of our cluster analysis (Kmeans method, see diagram 2), can be described in the following way in view of the coordinates of modalities and their location of the map.

(17)

• Diagram 1: The three groups identified

- Group 1 ("Economics cross-sectional"): Group related to conference database, in which theories originated more from the new institutionalism in economics (or mix), relying on quantitative techniques (second generation): surveys, experiments, with a cross-sectional perspective, an inter-organizational vision, often combining several industries.

- Group 2 ("Sociological longitudinal"): Group related to journals (clearly), in which neo-institutional theories of sociology dominate, following a longitudinal analysis, a structured qualitative method or other original methods (meta, hybrids) were used. The object of the research was not a specific technical concept, but IT on the whole or an organization IS. The perspective was often societal.

- Group 3 ("Atheoretical with mix approaches"): Very different from the other two groups, in particular from the methodological point of view (use of meta approaches, action-research or hybrids approaches). The object of action-research was a more specific technical con-cept, studied by means of a case study, applied to a single organization (intra-organizational), in a single industry, often by means of an unstructured qualitative data analysis. This research also rarely relied on a clear-cut, unique theoretical framework, but more on combinations.

We clearly, noticed that several variables did not really differentiate our groups, such as years of use of international comparison. Nonetheless, it seems that the years 2000 and 2007 were rather atypical (see diagram 2). One possible explanation is the transitional status of these years (where articles with various topics and perspective have been pub-lished, related to various groups).

4. Discussion: contributions, limitations and avenues for further research We see several contributions in this work.

(18)

First, a careful study of the modalities of instrumentation of a major theoretical perspective in IS (institutional and neo-institutional approaches) has been suggested. Our work con-firms the penetration and vitality of this research, which remains dominated by cross-sectional strategies , and a certain methodological orthodoxy Indeed, we saw few papers claiming an innovative methodological approach, like combinations of quantitative and qualitative techniques. Inter-cultural aspects remained under-explored (legitimating process of technologies were assumed to be universal, acultural), and ultimately, most researchers built middle range theories and concepts to make sense of empirical dynamics . This clearly epitomizes the meta-theoretical dimension of institutional perspectives, and their problem-atic instrumentation.

More indirectly, this work also illuminates some incoherencies. In continuation to Mignerat and Rivard (2005), it is striking to see the domination of cross-sectional, cross-sectional research (for conference papers). Conversely to the spirit of institutional approaches which valorises a longitudinal study of variables, the bulk of the empirical literature we found uses either instantaneous or short-term longitudinal approaches (this has been largely con-firmed by our reading of the papers). Surprisingly, longitudinal research retained a short-term perspective (several years). This view is not really coherent with institutionalist ambi-tions. Veblen (1898), in his study of the emergence of a leisure class, Kieser (1989) or Di-Maggio and Powell (1991) in their analysis of formal organizations, were interested in long term, even secular evolutions of organizations.

Lastly, the third contribution we see is methodological. This research has thus far been an opportunity to examine the different ways to instrument and operationalize very abstract theoretical frameworks and not to separate theoretical and methodological debates (both aspects are present in our thematic coding and canonical analysis). The concrete result is a set of criteria that enable to analyze instrumentation strategy of articles, i.e. the thematic dictionary. This set of criteria has been validated through a collective process before being used. The 85 articles have been analyzed through this grid. Beyond classic thematic and citation analysis, lexicometric analysis and canonical analysis have also been a way to map, in an innovative way, a stream of literature. The combining use of canonical and cluster analyses is not often used in IS literature and the results from this combination are interest-ing: we elaborated a set of groups (by also including a temporal variable) which are inter-esting to illuminate the structuration of empirical research drawing on institutional frame-works. Our work also suffers from several limitations.

First, a careful study of the modalities of instrumentation of a major theoretical perspective in IS (institutional and neo-institutional approaches) has been suggested. Our work con-firms the penetration and vitality of this research, which remains dominated by cross-sectional strategies13, and a certain methodological orthodoxy (we saw few papers claiming an innovative methodological approach, like combinations of quantitative and qualitative techniques). Inter-cultural aspects remained under-explored (legitimating process of tech-nologies were assumed to be universal, acultural), and ultimately, most researchers built middle range theories and concepts to make sense of empirical dynamics14. This clearly epitomizes the meta-theoretical dimension of institutional perspectives, and their problem-atic instrumentation.

13 Very generally for communications, and only for quantitative research in the case of journals. 14

(19)

Lastly, our work can be followed up by several avenues of research, whether for those in-terested in the study of institutional works, or more indirectly, those wishing to understand the structuration of the research field.

For the institutionalist, it is necessary to put the dynamic of organizations or populations of organizations into perspective over long period of time (de Vaujany, 2006). A detour to-wards historical methods (still rarely realised in IS) might be useful. Some works could thus deal with the structuration of IS and computer tools in a very old set of organizations like religious congregations or old national administrations. Methods and concepts of the "new history" may be helpful to understand isomorphism over very long period of time (see Le Goff, 2006).

For forthcoming research about the structuration of institutional research (in particular em-pirical studies), overviews on the way some precise variables have been operationalised (like "organizational fields" or "isomorphism") could also be an interesting new step. Finally, a cross analysis with the same literature in strategic management or human re-source management literature (where institutional frameworks also experience a certain success) could also be a way to shed light on interesting specificities of our field.

Lastly, our work can be followed up by several avenues of research, whether for those in-terested in the study of institutional works, or more indirectly, those wishing to understand the structuration of the research field.

For the institutionalist, it is necessary to put the dynamic of organizations or populations of organizations into perspective over long period of time (de Vaujany, 2006). A detour to-wards historical methods (still rarely realised in IS) might be useful. Some works could thus deal with the structuration of IS and computer tools in a very old set of organizations like religious congregations or old national administrations. Methods and concepts of the "new history" may be helpful to understand isomorphism over very long period of time (see Le Goff, 2006).

For forthcoming research about the structuration of institutional research (in particular em-pirical studies), overviews on the way some precise variables have been operationalised (like "organizational fields" or "isomorphism") could also be an interesting new step. Finally, a cross analysis with the same literature in strategic management or human re-source management literature (where institutional frameworks also experience a certain success) could also be a way to shed light on interesting specificities of our field.

References

Backhouse J, Hsu CW, and Silva L. "Circuits of power in creating de jure standards: shap-ing an international information systems security standard", MIS Quarterly, (30:4), 2006, pp. 13-38.

http://www.car.ua.edu/

Bardin L. L’analyse de contenu, 9e édition, Paris, Presses Universitaires de France, 1998, 291p.

Barley, S. et Tolbert PS. "Institutionalization and structuration: studying the links between action and institution", Organization studies, (18:1), 1997; pp. 93-117.

Barrett, M., Sahay, S., and Walsham, G. “Information Technology and Social Transforma-tion: GIS for Forestry Management in India”, The Information Society (17), 2001, pp. 5-20.

Bennett Thatcher J , Brower R.S., and Mason R. “ Organizational Fields and the Diffusion of Information Technologies Within and Across the Nonprofit and Public Sectors A

(20)

Pre-liminary Theory", The American Review of Public Administration, (36:4), 2006, pp. 437-454.

Bureau Sylvain "La professionnalisation des nouveaux métiers liés aux technologies de l'in-formation et de la communication : un déterminant dans les processus d'organisation d'une fonction ? Le cas des technologies web", Systèmes d'Information et Management, (11 :1), 2006, mars, pp. 5-22

Carton S., de Vaujany F.X. and Romeyer C. “Le modèle de la Vision Organisante: un essai d’instrumentation”, Systèmes d’Information et Management, (8:4), 2003, pp.3-29. Carton S., de Vaujany F.X. and Romeyer C. "Organizing vision and local IS practices: a

France-US comparison", Communications of Association for Information Systems, (19:11), 2007.

Chae B and Poole MS. "The surface of emergence in systems development: agency, institu-tions, and large-scale information systems", European Journal of Information Systems, (14:1), 2005, pp. 19–36.

Coase, R. “The Nature of the Firm” Econometrica, (4), 1937, pp. 386-405.

Commons, JR. "Institutional Economics", American Economic Review, (21), 1931, pp 648-657.

Commons, JR. The economics of collective action, Mac Millian, 1951, new edition in 1970.

Damsgaard, J., and Lyytinen, K. "The Role of Intermediating Institutions in the Diffusion of Electronic Data Interchange (EDI): How Industry Associations Intervened in Den-mark, Finland, and Hong Kong ", The Information Society, (17:3), 2001, pp. 195-210. Damsgaard, J., and Scheepers, R. “A Stage Model of Intranet Technology Implementation

and Management,” Proceedings of the Seventh European Conference on Information Systems, Copenhagen Business School, Copenhagen, Denmark, 23-25, June 1999. de Vaujany FX. "Conceptualizing IS archetypes through history: the case of the Roman

Cu-ria" ICIS conference, 2006, Milwaukee.

de Vaujany FX. "Capturing reflexivity modes in IS: a critical realist approach" Information. and organization, (18:1), 2008, pp. 51-72,

Di Maggio PJ. and Powell WW. "The iron cage revisited: institutional isomorphism and collective rationality", American Sociological Review, 48(2), 1983, pp. 147-160.

Di Maggio PJ. and Powell WW. "Introduction". in Powell WW et DiMaggio PJ (Eds.), The new institutionalism in organizational analysis, pp. 1-38. University of Chicago Press, Chicago, 1991.

Firth DR. “The organizing vision for customer relationship management”, ACIS 2001, pro-ceedings.

Giddens, A. The constitution of society, University of California Press, 1984. Gomez PY. Le gouvernement de l’entreprise, Interéditions, 270 p, 1996.

Limayem, M., and Hirt, S. G. "Force of habit and information system usage: Theory and initial validation", Journal of the Association for Information Systems, (4), 2003, pp. 65-97

Hodgson G. "Marching to promised land? Some doubts on the supposed theoretical and policy implications of critical realism", Workshop on realism and economics, King's College, Cambridge 26 April 1999.

Hodgson G. The Evolution of Institutional Economics : Agency, Structure and Darwinism in American Institutionalism, Routledge, 2004.

Jones, M. "Structuration theory" in Currie W. et Gallier B. (eds), Rethinking Information systems management, Oxford University Press, 1999.

Kieser, A. "Organizational, Institutional, and Societal Evolution: Medieval Craft Guilds and the Genesis of Formal Organizations", Administrative Science Quarterly, (34:4), 1989, pp. 540-564.

(21)

King J.L. Gurbaxani V., Kraemer K.L., Mc Farlan F.W., Raman K.S., and Yap C.S. “Insti-tutional factors in Information Technology Innovation” Information Systems Research, (15:2), 1994, pp. 136-169

Kling R. and Iacono S. "The institutional character of Computerized Information Systems", Office, Technology and People, (5:1), 1989, pp.5-27.

Lawson T. Economics and reality, Routledge, 1997.

Le Goff, J. (eds), La nouvelle histoire, Editions complexe, 1978.

Huberman, M. and Miles, MB. (Eds.) The Qualitative Researcher's Companion. Thousand Oaks: Sage, 2002.

Meyer J. W., Rowan B., "Institutional organizations: formal structure as myth and cere-mony," American Journal of Sociology, 83, 1977, 340-63.

Mignerat, M. and Rivard S. "Positioning the institutional perspective in information tech-nology research", Actes du congrès de l'ASAC, Toronto, May 2005.

Moe T.M. “The new economics of organizations”, American Journal of Political Science, (28), 1984, pp. 739-777.

Montealegre R. "A temporal Model of Institutional interventions for Information Technol-ogy Adoption in Less-Developed Countries", Journal of Management Information Sys-tems, (16:1), 1999, pp. 207-232.

North, D.C. Institutions, institutional change and economic performance, Cambridge Uni-versity Press, 1990.

Oliver, C. “Strategic Responses to Institutional Processes”, Academy of Management Re-view, (16), 1991, pp. 145-179.

Orlikowski, W., and Barley, S. "Technology and Institutions: What Can Research on In-formation Technology and Research on Organizations Learn from Each Other?", MIS Quarterly, (25:2), 2001, pp. 145-165.

Robey D., Boudreau M-C “Accounting for the Contradictory Organizational Consequences of Information Technology: Theoretical Directions and Methodological Implications”, Information Systems Research, (10:2), 1999, pp 167-185.

Rojot, J. Théorie des organisations, Editions ESKA, second edition, 2005. Scott, W. Institutions and Organizations. Thousand Oaks, CA: Sage, 2001.

Scott W.R. “Institutional Theory : contributing to a research Program” Chapter in Great Minds in management : the process of theory Development, K.G. Smith M.A Hitt eds Oxford UK:Oxford university Press, 2004.

Searle, JR. "What is an institution ?", Journal of Institutional Economics, (1:1) , 2005, pp. 1-22.

Selznick, PK. Leadership in administrations, Peterson adn co, Avanston, 1957. Selznick, PK. Law, Society and Industrial Justice, Transaction books, 1969.

Silva L., Figueroa E. "Institutional intervention and the expansion of ICTs in Latin Amer-ica, the case of Chile", Information Technology and People, (5:1), 2002, pp. 8-25. Swanson EB. and Ramiller NC. “Innovating mindfully with information technology”, MIS

Quarterly, (28:4), 2002, pp. 553-583.

Swanson EB. and Ramiller NC. “The organizing vision in information systems innovation”, Organization Science, (8:5), 1997, pp 458-474.

Tennenhaus, M. Méthodes quantitatives de recherche en gestion, Dunod, 1993.

Tornatzky Fleischer M. The processes of Technological Innovation, Lexington MA Lex-ington books, 1990.

Valette-Florence P. “A causal analysis of means-end hierarchies in a cross-cultural context : methodological refinements” Journal of business Research 42, 161-166

(22)

Walsham, G. Interpreting Information Systems in Organizations, John-Wiley, Chichester, England, 1993.

Wang P. “What drives waves in Information Systems? The Organizing Vision perspective”, Twenty-second International Conference of Information Systems, ICIS proceedings, pp. 410-416, 2001.

Wang P. and Swanson, EB. "Launching professional services automation: Institutional en-trepreneurship for information technology innovations", Information and Organization, (17:2), 2007, pp. 59-88.

Weber, RP. Basic content analysis, Sage, 1990.

Williamson, OE.Market and hierarchies. Analysis and antitrust implications, The Free Press, 1975.

Zucker, LG. "Institutional theories of organization", American Review of Sociology, (13),

1987, pp. 443-64.

APPENDICES

APPENDIX 1 : Main descriptive analysis of our thematic

cod-ing

Code Frequency for the article database (ABI and EBSCO)

Frequency for the Communication database (ICIS)

METH-CAS (Case studies) 51% 24.3%

METH-ENQ (Surveys) 30.6% 62.2%

METH-EXP (Experiments) 4.1% 8.1%

METH-HYB (Hybrid methods) 6.1% 0%

METH-META (Meta-analyses) 8.2% 5.4%

METH-RECHAC (Action research) 0% 0%

INTO-ORG (Inter-organizational perspec-tives)

40.8% 51.4%

INTRA-ORG (Intra-organizational per-spectives)

28.6% 18.9%

SOC-PERS (Societal perspectives) 30.6% 29.7%

CONC-TECH (Technical concepts) 38.8% 24.3%

IS-GAL (IS or IT) 61.2% 75.7%

Year 1999 4.1% 16.2% Year 2000 4.1% 10.8% Year 2001 10.2% 8.1% Year 2002 8.2% 8.1% Year 2003 10.2% 13.5% Year 2004 10.2% 13.5% Year 2005 14.3% 10.8% Year 2006 28.6% 18.9%

RECH-LONG (Longitudinal research) 63.3% 29.7%

RECH-TRANS (Cross-sectional research) 36.7% 70.3% RECH-QUALINS (Unstructured

qualita-tive research)

24.5% 13.5%

(RECH-QUALI (Structured qualitative research)

(23)

QUANT1 (First generation quantitative research)

24.5% 24.3%

QUANT2 (Second generation quantitative research)

14.3% 48.6%

INTER (Intercultural) 16.3 10.8%

MONO-CULT (Monocultural) 83.7% 89.2%

INT-SEC (Inter-sectorial) 51% 56.8%

INTRA-AD (Intra-sectorial administra-tion) 16.3% 5.4% INTRA-SER (Intra-service) 18.4% 24.3% INTRA-IND (Intra-industry) 14.3% 13.5% ACCT-AUT (Partly-other) 14.3% 18.9% ACCT-INST (Partly-institutional) 4.1% 0%

ACCT-NEOMIX (Partly-new institutional economic)

2% 27%

PPT-NEOMIX (Partly-new institutional mix)

4.1% 2.7%

PPT-NEOECO (Partly-new institutional in sociology)

12.2% 18.9%

PPT-AUT (Mainly-other) 6.1% 2.7%

PPT-INST (Mainly institutional) 2% 0%

PPT-NEOECO (Mainly-new institutional economics)

6.1% 16.2%

PPT-NEOMIX (Mainly-new institutional-ism mix)

4.1% 2.7%

PPT-NEOSOC (Mainly new institutional-ism in sociology)

44.9% 10.8%

JOURN-AUT (Other journals) 22.4% Not relevant

JOURN-MANT (Journals in management) 6.1% Not relevant

JOURN-IS (Jiournals in MIS) 71.4% Not relevant

APPENDIX 2 : LEXICOMETRIC ANALYSIS (EXTRACT)

For every years (on the consolidated database)

information 31 systems 15 technology 13 Institutional 12 e 9 IT 9 organizational 9 development 8 perspective 8

(24)

24

A

P

P

E

N

D

IX

3

: C

L

U

S

T

E

R

A

N

A

L

Y

S

IS

D e n d ogr am ( w it h n = 37) w it h War d 's m e th od . D e n d ro g ra m m e d e 3 7 O b s. M é th . d e W a rd D is t. E u c lid ie n n e s

ACCT neo eco colloques Quanti 2ème générati PPT neo eco ACCT neo socio 1999 Expérimentations 2005 Inter-organisationne 2003 transversal Enquête Autres revues (manag Qualistructur PPT neo socio Management des SI longitudinal 2007 Autres (AUT + INST) 2002 NON SI ou NTIC en généra Sociétal 2006 Inter-sectoriel Quanti première gén Autres (Meta, RA, Hy 2000 Quali non-structur Mixtes (ACCT + PPT) OUI intra-sectoriel Concept technique 2004 2001 intra-orga Mono et cas 0 1 2 Dist. Agrégation3 4 5 6 7 8 9

(25)

Test F (ANOVA)

ANOVA

Cluster Error

F Sig.

Mean Square df Mean Square df

DIM1 3,956 2 ,125 34 31,734 ,000

DIM2 3,532 2 ,202 34 17,482 ,000

The F tests should be used only for descriptive purposes because the clusters have been chosen to maxi-mize the differences among cases in different clusters. The observed significance levels are not corrected for this and thus cannot be interpreted as tests of the hypothesis that the cluster means are equal.

Discriminant analysis for each group

Classification Results(b,c)

Cluster Number of Case Predicted Group Membership

T o t a l 1 2 3 Original Count 1 11 0 0 11 2 0 17 1 18 3 0 0 8 8 % 1 100,0 ,0 ,0 100,0 2 ,0 94,4 5,6 100,0 3 ,0 ,0 100,0 100,0 Cross-validated(a) Count 1 11 0 0 11 2 0 17 1 18 3 0 0 8 8 % 1 100,0 ,0 ,0 100,0 2 ,0 94,4 5,6 100,0 3 ,0 ,0 100,0 100,0

a Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the functions derived from all cases other than that case.

b 97,3% of original grouped cases correctly classified. c 97,3% of cross-validated grouped cases correctly classified.

(26)

APPENDIX 4 : PRESENTATION OF THE RECODING

PROCESS BEFORE OVERALLS ANALYSIS

10 variables, 37 modalities

VARIABLES MODALITIES RECODED

MODALITIES 1 Methodological

ap-proaches (apprométho)

Action-research, survey, case study, experiment, meta-analysis (i.e. analysis of a set of previous em-pirical research), hybrid approach (combination of the previous ones).

1. Case study 2. survey 3. experiment

4. others (Hybrid and meta nalysis)

2 Level of analysis (them) Societal, inter-organisational, intra-organisational. 5. cross-organizations 6. intra-organizations 7. societal 3 Period of publication (Date) 8. 1999 9. 2000 10. 2001 11. 2002 12. 2003 13. 2004 14. 2005 15. 2006 16. 2007 4 Target (Géné/spé) Specific technical concept,

Information technologies or Information Systems on the whole

17. Specific IT concept 18. IT or IS on the whole

5 Nature of data collection (Datcoll)

Cross-sectionnal (i.e. punctual) or longitudinal

19. cross-sectional 20. longitudinal

6 Nature of data treatment (Nattrait)

- First generation of quan-titative techniques: simple or multiple linear regres-sion, descriptive tests…; - Second generation quan-titative techniques: struc-tural equation modelling (like LISREL or PLS), factorial analysis, depend-ency network diagrams, logistical regression, game theory techniques… - Unstructured qualitative methods (no coding and thematic dictionary was mentioned);

- Structured qualitative methods (with a coding and treatment of the cod-ing)

21. unstructured qualitative methods 22. structured qualitative methods 23. first generation quantitative methods 24. second generation quantitative methods

(27)

(Appcompint) 26. yes 8 Sectorial approach (Appcompsectorielle) Intra-administration indus-try, intra-manufactury industry, intra-agrarian industry, intra-service industry, cross-sectorial 27. Cross-sectorial 28. One sector 9 Theoretical framework instru-mented/implemented (cadtheo)

Mainly or partly institu-tional, Mainly or partly neo-institutional in the field of sociology, Mainly or partly institutional in the field of economics, mixt or with an implicit positioning.

29. Partly neo institutional (economics) 30. partly neo institutional (sociology) 31. mainly neo institutional (economics) 32. mainly neo institutional (sociology) 33. Others (other institutional frameworks) 34. Mix (weak references of institutional

frameworks)

10 Types of publications management, IS manage-ment or others.

35. management revues except IS ones 36. IS revues

Figure

Table 5: Number of modalities in each cluster

Références

Documents relatifs

As we mentioned before, if we only look for a short time solution, the fully explicit scheme for the nonlinear convection term may be analyzed accordingly as well.. Moreover,

In order to create the deformed vaults of the deambulatory (vaults around the main circular vault of the choir), we have extended the parametric circular vault:

Category 3 reflects students’ pure prototypical images because they select only all proto- typical examples of trapezoids according to exclusive definition even they know

decomposition strategy described in Section 2 (see Figure (13)) along with Algorithm (4.1)). This idea is similar to the concept of nonlinear relocalization developed in [5, 27],

Our main hypotheses are (H1) agricultural practices affect soil nema- tode communities in different ways (e.g. tillage affects large nematodes, while managing plant diversity

A Critical Analysis of Husserlian Phenomenology in IS Research and IS Development: The Case of Soft

In terms of empirical results, we have provided an estimation of the average value of ecosystem services provided by lakes (between 106 and 140 USD$2010 per respondent per year for

We establish recursion relations for the generating functions of the average numbers of (variable or color) assignments at a given height in the search tree, which allow us to