Do Organizational Memory and Information Technology Interact to Affect Organizational Information Needs and Provision

Do organizational memory and information technology interact to affect organizational information needs and provision?1,2

Mohamed El Louadi

Institut Supérieur de Gestion University of Tunis - Tunisia

and

Imen Tounsi

Hautes Etudes Commerciales University of Sousse - Tunisia

August 2007

Abstract The results reported in this paper were obtained through the study of 43 work-units belonging to five business firms in a large group of Tunisian companies. Using the declarative, procedural and judgmental dimensions of organizational memory, interaction effects were detected between IT and organizational memory. Analyses of the data support the view that storage technologies interact with declarative memory contributing to reduce organizational information needs. But, contrary to expectations, storage technologies interact negatively with declarative memory on information provision. Furthermore, network technologies interact negatively with judgmental memory on information provision. These results raise issues that have been rarely encountered in the literature. Keywords: Organizational Memory, Information Technology, Information Needs, Information Provision, Tunisia

1 An early draft of this paper has been presented at the First International e-Business Conference, held in Hammamet (Tunisia) on June 23-25, 2005. 2 This paper is being reviewed for publication in an international peer-reviewed journal. Do not cite without permission from the first author.

1

Do organizational memory and information technology interact to affect organizational information needs and provision?

Abstract

The results reported in this paper were obtained through the study of 43 work-units belonging to five business firms in a large group of Tunisian companies. Using the declarative, procedural and judgmental dimensions of organizational memory, interaction effects were detected between IT and organizational memory. Analyses of the data support the view that storage technologies interact with declarative memory contributing to reduce organizational information needs. But, contrary to expectations, storage technologies interact negatively with declarative memory on information provision. Furthermore, network technologies interact negatively with judgmental memory on information provision. These results raise issues that have been rarely encountered in the literature. Keywords: Organizational Memory, Information Technology, Information Needs, Information Provision, Tunisia

Introduction

Internal organization’s information-processing capacity has almost always been related to structural and technological mechanisms (Girod, 1996; Huber, 1990; Cummings and Cross, 2003; Walsham and Barrett, 2005). Quicker and easier access to external information and knowledge is becoming increasingly more important to organizations. Since the 1990s the accumulation of data and information in organizations files and databases led researchers’ interest to the study of organization memory too (Souren et al., 2004). Organization memory has become an additional mechanism allowing organizations to store, process, and disseminate information with the aim of speeding up organizational learning (Child and Rodrigues, 2001; Cohen, 1991).

Unfortunately, there has been very little empirical research on organizational memory and its relationship with the amounts of information needs and provision to date, despite the increasing realization that organizations are becoming more knowledge-based, relying not only on their computerized information systems but also on information and feedback from workers, customers, and partners. Even so, the knowledge-based company depends increasingly on knowledge, i.e., patents, processes, management skills, old-fashioned experience, past failures and successes (Levitt and March, 1988; Walsh and Ungson, 1991). This is especially true in the case of organizations whose members are specialized in some particular task and where learning occurs essentially by doing. Such companies have managers who need to access more information about more aspects of the business, from more parts of the organization, and in shorter time spans (Huber, 1984).

While organizations develop uncertainty-reduction mechanisms by increasing their information-processing capacities favoring the acquisition, creation, and flow of new information (Galbraith, 1977), some other mechanisms are needed for the processing of information that the organization already possesses in order to foster learning. Some of the knowledge needed is accumulated in databases and archived documentation but the most important part lies in people’s heads (Nonaka, 1995). An organization’s knowledge capacity is therefore a function of the sum of everything everybody in the organization knows (Aschcraft, 1994; Walsh and Ungson, 1991).

Organizations do not need knowledge about past events only. They also need knowledge about past and present happenings and experiences. Learning cannot be dissociated from memory (Lehner et al., 1998, Walsh and Ungson, 1991); it is “…a process that involves the discovery,

2

retention, and exploitation of stored knowledge” (Moorman and Miner 1998, p. 705). Mechanisms related to accumulated information about past and present events are collectively known as the organizational memory (Walsh and Ungson, 1991; Stein and Swass, 1995).

This research endeavors to explore relationships between information technology (IT) and organizational memory, and their combined effect on the provision of information. It aims at answering the question as to whether organizational memory is associated with the degree of information availability and, if so, whether it is augmented by IT in providing needed information.

We have organized this paper as follows. First, we present a definition of organizational memory and three of its commonly used dimensions: declarative memory, procedural memory, and judgmental memory. Second, we present a conceptual overview of the concepts. Then we introduce the method and data collection procedure. In the following section, we present the measures we used. Following this, we present the results of the tests of the relationships between the variables which we also discuss then we conclude the paper with the limitations of the study and propositions for future research.

Organizational Memory

Building on Walsh and Ungson (1991), Stein and Swass (1995) define organizational memory as “the means by which knowledge from the past is brought to bear on present activities” (p. 22).

Organizational memory has been conceptualized theoretically in several different ways. Ashcraft (1994), for example, classifies the different types of memories as episodic (knowledge of events personally experienced by people), semantic (factual knowledge), or procedural (skills learned, or as Markus (2001) calls it, “best practice”). Girod (1995) defines organizational memory as a set of competencies associated with beliefs and declarative and procedural knowledge borne out of inter and intra-organizational structural arrangements.

Synthesizing several literatures, Girod (1995, 1996) distinguishes between three memory categories. The first, called “declarative memory”, is a mostly explicit memory containing knowledge accumulated in human memories about facts, things and events. The second, “procedural memory”, is mostly tacit and contains knowledge about how to do things. And the third, “judgmental memory”, contains memories related to the individual’s experience.

Taken at the organizational, collective level, declarative and procedural memories are well known constructs in other management literatures (Anderson, 1983, Moorman and Miner, 1998) and to cognitive psychologists (Chase and Simon, 1973; Singley and Anderson, 1989). For that reason, organizational memory will be taken as composed of its three categories: declarative, procedural, and judgmental, which, according to Girod (1995) can also be conceptualized at the collective level as well.

Declarative Memory – The Know What Declarative memory is the sum of technical, scientific, and administrative knowledge detained by organizational members which is related to their work and which, therefore, should be made available to the organization (Girod, 1996). It consists of explicit knowledge, i.e., facts, propositions, events, situations, etc. (Moorman and Miner, 1998). Thus what individuals learn in school about the operation of machinery and which they bring into the organization upon their hiring is declarative knowledge. This knowledge can be stored in the individuals’ memory or in archives, in the form of how-tos that other individuals can tap into. Girod considers that the memory containing declarative knowledge is not limited to individuals. It can be derived from the interaction of two or more people. In this case, it is considered collective declarative memory. It can be formalized and stored in databases for example (centralized collective declarative memory) or informal, implicit, unrecorded, and volatile (decentralized collective declarative memory).

At the collective level, declarative memory can be likened to what Walsh (1995) terms as “shared information”. Moorman and Miner (1998) note that a key characteristic of declarative

3

memory is the diversity of uses to which it can be put thus providing the basis for transfer of the same knowledge between different applications.

Procedural Memory – The Know How Procedural memory is composed of the know-how that individuals apply in their day-to-day work-life (Cohen and Bacdayan, 1994). It is skill and action knowledge (Moorman and Miner, 1998). Unlike declarative knowledge, which is stored in declarative memories, procedural knowledge is the result of an individual’s personal learning experience. A learning organization tries to transform this knowledge into organizational knowledge to make it available to everyone (Girod, 1996). Routine knowledge transformed into procedural knowledge is stored into rules and procedures (Cohen and Bacdayan, 1994) and is categorized as centralized collective procedural memory by Girod (1995). Moorman and Miner (1998) note that procedural memory generally becomes “automatic” or that it is accessible “unconsciously” (p. 708). Most often tacit, procedural memory can, however, be made explicit when it is documented and circulated in the shape of procedures, for example.

Judgmental Memory – The Know Why Judgmental memory is the individuals’ tendency to interpret information, events, and knowledge to derive proper action. It is memory about why things are done. Judgmental memory, also called “rationale memory” by Moran and Carroll (1996), relies forcibly on individuals’ own experience and knowledge. It is what makes the opinion of an expert more valuable than a layperson’s. It is also the type of memory that suffers the most from personnel turnover because it is very hard to extract and record. At the collective level, judgmental memory constitutes the sum of all organizational members’ judgmental memories and the unique way that an organization senses and interprets events.

Organizational Memory and IT

Memory is increasingly being regarded as a resource, like labor and capital. In that, it is on the same path as information. In the information-processing view, an organization is viewed as an ensemble of information-processors (humans, machines, networks, etc.) matching information and knowledge acquired from the outside with information and knowledge accumulated in order to improve decision-making and effectiveness.

In an ethnographic study of a help line group, Ackerman and Halverson (2000) demonstrated the theoretical limitations of the metaphor of a single organizational memory. They contend that there is a supra-individual memory, maybe even several memories, complexly distributed among several people and technologies. Walsh and Ungson (1991) consider memory to include not only people but also archives, organizational procedures, organizational structures, and culture, but not information technology. In their model, Walsh and Ungson (1991) have categorized organizational memory as being spread over six organizational contexts: individuals (people’s heads), culture, organizational transformations, structures, organizational ecology, and external archives. Ackerman (1996) argues that information repositories such as corporate manuals, databases, filing systems, and even anecdotal stories could also be included in the construct of organizational memory.

Yet, one of the objectives of IT is to make the required information easily accessible. IT plays an important role in disseminating information and in facilitating communication within organizations (Huber, 1990; Walsham, 2001). Like Huber (1990 p. 65), our hypothesis is that use of IT should lead to more information availability and greater information accessibility.

It can also be easily argued that IT represents a way for organizations to augment their organizational memory (Ackerman, 1996; Lehner et al., 1998; Moorman and Miner, 1998). As a consequence, IT would become embedded in a broader conceptualization of organizational memory as contended by Watson (1996) who considers organizational memory to be essentially

4

technology driven, a comprehensive enterprise-wide database supported by multimedia, imaging, and electronic archiving and document management systems.

Conversely, a distinction can be made between memory that is based on IT, also called organizational memory systems (OMS) (Ackerman, 1996; Lehner et al., 1998), and organizational memory information systems (OMIS) (Wijnhoven, 1999). While Walsh and Ungson (1991) did not count computer files and databases among their components of organizational memory, Wijnhoven (1999), for example, includes in it individuals, culture, transformation, structure, ecology, external environment, computer-based information systems, and non-IT based records and files. Consequently, there would be two types of memories, one that is based on computer and communication technologies and another which is based on non IT-based mechanisms, both hypothesized to contribute positively to organizational information-processing capacity. As noted by Ackerman (1996):

“Organizational memory systems (OMS) offer the possibility that computer systems can better serve the information storage and retrieval needs of an organization’s memory than can present technical and social methods.” (Ackerman 1996, p. 10).

Non IT-based memory, especially that which lies in people’s heads has the advantage of being available and relatively more accessible within a known context, but it suffers from poor retention and is volatile because it is prone to leaving the organization once the individual holding it leaves (Stein and Zwass, 1995) causing what Borghoff and Pareschi (1997) and Kransdorff (1998) call “corporate amnesia”. Thus, non-computer based memory can suffer corruption, decay in both detail and accuracy or even erasure with time. Knowledge in people’s heads comes and goes as people come and go (Nonaka, 1995).

IT-based memory, on the other hand, is quasi-permanent; knowledge is logged, indexed, secured, and organized in such a way that it becomes accessible. However, IT-based memory suffers from limitations that have been pointed out by Ackerman (1996) such as the impossibility of having a unified organizational memory reflected in incomplete, fragmented and/or overlapping databases and files. Furthermore, IT often lacks the capability of providing necessary clues in light of the context in which memory is found most useful.

IT contributes to organizational memory in at least two ways: by making accumulated knowledge accessible to organization members or by making individuals with knowledge known (Ackerman, 1996; Jennex, 2006). However, there is still no agreement as to which technologies best support this memory and Lehner et al. (1998) suggest that organizational memory cannot be served by a single technology. Cross and Baird (2000) contend that it is not enough to accumulate information and knowledge in electronic repositories and that organizational members rely upon a network of relationships for information and advice. According to Cross and Baird (2000), people take advantage of databases only when more trained colleagues point them to a specific location in the database where they can find the information that they seek. For Cross and Baird, IT is only one form of memory that organizational members tap into when solving problems and its use is therefore limited.

In this paper, we consider, like others (Jennex, 2006; Huber, 1991; Lehner et al., 1998; Stein and Zwass, 1995), that IT is also used to support organizational memory.

Setting The study was conducted in Tunisia in 2002. Tunisia is a small 10 million-people North-African country. In 1995, it became the first Maghreb and Arab country to sign an association agreement with the European Union. Since then, Tunisia’s economy has grown by over 4.5% annually. Three years later, it ratified the European Association Agreement which calls for the gradual

5

removal of all tariffs on industrial goods by 2008. Today, Tunisia is seriously preparing its industrial sector, composed of more than 10,800 firms of more than 10 employees, to international competition through a $2.5 billion competitive restructuring program, launched in 1995 and dubbed ''The Way to Competition''. More than 2,360 Tunisian firms are totally exporting, and this number is increasing. These firms face the challenge to be as competitive as their European counterparts, maybe even as their other Asian and American partners.

Conscious of these implications, and encouraged by a set of governmental policies, the computer and telecommunication sectors have greatly contributed in making Tunisia an ideal regional telecommunications hub with access to Europe, the Middle East and Africa. More than TD 1779 million were invested in IT as part of the 10th Plan (2002-2006)3 and TD 3107 million are planned in the 11th five-year Economic and Social Development Plan (2007-2011).

According to an International Herald Tribune article (Stratte-McClure, 2000), “Advances in telecommunications, the Internet and electronic commerce are appropriate symbols of Tunisia's industrial, social and economic strides”. Having been the first African and Arab country to connect to the Internet in 1991, when there were only thirty-one countries connected (ITU, 2003), Tunisia now boasts more than 1 446 000 Internet users (roughly 14.5% of the population)

4.

In 2005, Tunisia was ranked 31st according to the Network Readiness Index developed by the World Economic Forum (WEF), up from the 40th position it was in the previous year (Forbes, 2005). It was ranked No.1 in Africa and in the Arab world for economic competitiveness in the 2007 WEF Global Competitiveness Report. Based on an assessment of the economies of 128 countries, the report ranks Tunisia the 29th most competitive country in the world. Tunisia was also ranked by the WEF as the first Arab and African country and 35th in the world in the field of global information technology5. Tunisia is currently working out an overall IT strategy for the 2007-2011 period with the support of the World Bank, with an aim of reaching an IT infrastructure equivalent to that of the developed countries by 2011.

However, even though Tunisia has often been touted by the International Monetary Fund (IMF) and the World Bank as models for its emerging economies, IT, the main driver for its integration into world economy, seems to have been mostly adopted by the industry for clerical purposes. With a few exceptions, like the group of companies we studied.

Data Collection

The group of firms in which we conducted our study was created in 1967, 11 years after Tunisia’s independence. In its beginnings, it was specialized in commercializing meat products such as chicken and turkey. In three decades, it expanded into a Holding composed of 30 firms organized in a matrix form and employing more than 6.000 people. Today, 95% of the Holding’s products are either new in Tunisia (such as industrial ceramic) or manufactured using new technologies. It exports some of its products to more than 35 countries. Highly modernized, having complied with most ISO 9000 and restructuring programs, the group is reputed for its sense of organization, processes and procedures, overseen by a central audit structure in headquarters.

It was that central audit unit that selected the companies as well as of the work-units within them which were to participate because the study was incorporated into a larger scale survey which was being made simultaneously. The only condition was that the questionnaire, written in 3 See The Tenth Development Plan, page 130, http://www.tunisieinfo.com/10plan-fr.pdf, last accessed 9 April 2005 (available in French only). 4 As of this writing, March 2007, see the Tunisian Ministry of Technologies and Communication, http://www.infocom.tn/index.php?id=163 and http://www.infocom.tn/uploads/media/Statistiques_internet_fin__Mars_2007_02.xls, last accessed 16 August 2007. 5 See http://www.weforum.org/pdf/gitr/rankings2007.pdf, last accessed 23 August 2007.

6

French, that we used for data collection be handed to the same units across firms. The firms were geographically dispersed in four cities. The units in each firm were: (1) Production, (2) Commercial, (3) Maintenance, (4) Accounting, (5) Finance, (6) Quality Assurance, (7) MIS, (8) Purchasing, and (9) Administrative and HR. The major focus of this study is on organizations work-units’ memory and information needs and provision. We therefore concentrated on collecting data from various work- units in five firms all belonging to the same group of companies.

Because the use of multiple informants is ideal while measuring organization-level constructs (Bruggen et al., 2002), we polled several informants from each work-unit. While the use of multiple informants from each work-unit is preferable, the single informant approach allows for a larger number of work-units to be surveyed. In the face of resource constraints, we opted for the latter approach. Consequently, we took the precaution of wording our questionnaire so as to make clear that the entire organization is to be considered6.

Our starting random sample of 45 (5 firms x 9 units) was reduced to 43 because two work-units did not respond in time or produced unusable answers. Forty three responses will consequently be used for analysis in the remainder of this paper (Table I).

Table I. Sample Characteristics. Main activity No. of Work-units #. employees (FTE) 1 Animal food industry 9 156 2 Animal food industry 9 48 3 Food industry 9 157 4 Food industry 9 324 5 Manufacturing 7 288 Total 43

The Measures

Measures of organizational memory, IT penetration and information needs and provision were used for this study. These measures are explained below.

Organizational Memory A measure composed of 15 items was specially designed for this study to measure organizational memory. Factor analysis yielded a three factor solution explaining 64% of the variance. Coinciding with Girod’s (1995) three dimensions, the factors were labeled (1) Procedural Memory, (2) Judgmental Memory, and (3) Declarative Memory (see Table II).

6 We refer the interested reader to Hage and Aiken's (1967) discussion on this subject and notably to footnotes 7 and 8 on page 76.

7

Table II. Item-to factor loadings for the Organizational Memory Variable. Procedural

Memory Judgmental

Memory Declarative

Memory Existence of documentation concerning past successful projects

.81

Existence of documentation concerning past failed projectsa .79 Existence of a procedure assuring the replacement of a leaving manager by another

.79

Number of qualified or specialized members in the unit .48 The extent to which experimented unit members are solicited for help and advice

.79

The extent to which help is available when an administrative problem occurs within the unit

.76

The extent to which help is available when a technical problem occurs within the unit

.71

The extent to which help is available when a social/relational problem occurs within the unit

.71

Frequency of individual activity reports production .95 Frequency of group activity reports production .92 Frequency of meetings within the unit .66 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with KaiserNormalization. a Reverse scored Amount of Information Needs and Provision Information needs and provision of any organization can be classified as internal or external. Internal information is generated within the organization to satisfy its working members and decision-making functions. It pertains to the organization’s internal operations. It supports decisions typically made by operations managers with short time horizons. External information is information that is of interest for the organization but which originates in its environment. It pertains to the organization’s customers, competitors, etc. (Ewusi-Mensah, 1981).

Respondents were asked to rate, on a scale from 1 (not at all) to 5 (a lot), the extent of availability of 10 categories of external information and 9 categories of internal external information. These measures were developed by El Louadi (1996) and their reliability and dimensionality were later further reassessed (El Louadi, 1998).

Degree of IT Adoption Information technology refers to the technologies and applications which combine the information processing, storage and transmission powers of computers with the distance-transmission capabilities of telecommunication technologies (Bakopoulos, 1985). It also complements and supports knowledge processing (Tuomi, 2000). From our previous discussion, IT can also be conceptualized as a mechanism helping to augment organizational memory.

Some technologies, such as electronic filing cabinets, document archiving systems, relational databases, Intranet, groupware, data warehouses, navigators’ information archiving with bookmarks, etc. have more of a memory focus than others. Unfortunately, there is still little evidence supporting the preeminence of some technologies over others with regards to organizational memory.

Shortly after its creation, the group in which we conducted our study was using a centrally controlled IT infrastructure with several systems and computer centers disseminated across the units. In the mean time, the group had diversified its activities to include eggs, ice cream, margarine, yogurt, wood, refrigerators, computer assembly, mufflers, tourism, etc. In 1991, it shifted towards clients-server computing progressively harmonizing its information systems until it adopted an ERP in 1998. In less than five years, it moved from unified transaction-processing

8

to reporting to on-line analytical processing thus acquiring and fully exploiting a huge data warehouse fed by the many data marts scattered over the units. As a result, the group is probably one of the most, if not the most, developed IS shop in the Tunisian industry today. It uses electronic mail extensively (1.200 accounts at the time of the study) and hires its employees through its Web site.

For this reason, we included all technologies available in the business units we surveyed.

In the questionnaire, we asked questions concerning the degree of adoption of IT using a 26-item list of information technologies. The respondents were asked to rate each technology on a scale from 1 (technology not operational) to 5 (operational on a widespread basis). Factor analysis with varimax (orthogonal) rotation and Kaiser normalization was then performed with the purpose of generating factors that are easily interpretable. A four -factor solution emerged which explains 66% of the scale’s variance. As shown in Table III, the factors were labeled (1) Software Penetration, (2) Storage Penetration, (3) Network Penetration, and (4) Communication Penetration.

Table III. Item-to-factor loadings for the IT Variable.

Software Penetration

Storage Penetration

Network Penetration

CommunicationPenetration

Accounting Software PC Software Inventory Mgmt. Software

.79

.75

.69

Personal Information Managers Personal Computers Databases (Access) Data warehouse

.78

.77

.59

.47

Fax Voice Mail Wide Area Networks

.74 .74 .69

External (Internet) e-mail Internal (Intranet) e-mail .85

.53 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization.

Analyses

Prior to analysis, the IT (software, storage, network, and communication) variables, the memory (procedural, judgmental, and declarative), and the information provision (external and internal) and information needs (external and internal) variables were examined through various SPSS programs for accuracy of data entry, missing values, and normality and multivariate analyses.

Because of the small size of our sample, we conducted tests of normality of variables following Tabachnick and Fidell (1989) recommendations by using descriptive programs for skewness. We found moderate negative skewness for the external information needs variable (skewness=-1.26). We corrected for skewness by computing the square root of the reflect of that variable. This reduced skewness considerably7.

We calculated the Shapiro-Wilk statistic W (Shapiro and Wilk, 1965), which has become standard for small samples with 50 or fewer observations (Royston, 1982). The Shapiro-Wilk test revealed that the hypothesis of non-normality could not be rejected for 4 out of 7 independent variables and 2 out of four dependent variables (W varying between .84 and .94 significant at 7 The reflect of a variable is obtained by deducting the scores from the largest value incremented by Tabachnick and Fidell (1989). Because the scale on which the External Information Needs variable was measured ranges from 1 to 5, we used the following transformation formula: transf=sqrt(6-x).

9

p=.024 or less).

We also looked for multivariate outliers. The tests revealed the existence of 11 multivariate outliers for which the Mahalanobis distance was greater than 3. Given the size of the data set, we were reluctant to eliminate them. However, we flagged the outliers with a dummy variable (see Tabachnick and Fidell 1989, p. 69) and applied the Shapiro-Wilk test on the remaining observations (the reduced sample). Thus, there were only three independent variables for which normality remained an issue (W≥.880; p≤.015).

Though the unit of analysis is the business unit, not work-units or individuals, aggregating the data within work-units could present methodological issues for which there are yet no satisfactory solutions if members within a work-unit perceive the dimensions studied here differently. To test the assumption that participants from each work-unit responded similarly, a series of ANOVAs was performed (Dixon and Massey, 1969). Because we observed large variances among individuals making up the groups and the way they are responding to the variables, the decision was not to pool individual responses to represent groups.

Table IV summarizes the statistical properties of the variables used in this study.

Table IV. Statistical Properties of the Variables of the Study. Min Max Mean Std-Dev α Procedural Memory 1.00 4.75 3.32 .97 .75 Declarative Memory 1.00 5.00 2.97 1.04 .76 Judgmental Memory 2.00 5.00 3.65 .81 .83 Software Penetration 1.00 5.00 4.12 1.11 .67 Storage Penetration 1.40 5.00 3.97 .94 .70 Network Penetration 1.00 5.00 3.10 1.29 .63 Communication Penetration 1.00 5.00 4.31 .95 .65 Internal Information Needs 2.00 5.00 3.89 .84 .88 Internal Information Provision 1.00 4.44 3.38 .81 .87 External Information Needs 1.00 5.00 3.63 .93 .89 External Information Provision 1.00 4.00 2.73 .73 .86

T-tests revealed that though the need for internal information is not significantly greater than the need for external information, the provision of internal information is significantly greater than the provision of external information (t=5.59, 42 df, p<.0001) (see Table IV and Figure 1). Furthermore, the need for information is significantly greater than provision whether it is internal (t=4.06, 42 df, p<.0001) or external (t=6.15, 42 df, p<.0001). The least we can infer from this result is that work-units in this sample do not suffer from information overload but rather from information poverty (Simon, 1982).

10

Externalinformationprovision

2.73

Internalinformationprovision

3.38

Externalinformation

needs3.63

Internalinformation

needs3.89 3.06 3.75

Meaninformationprovision

Meaninformation

needs

Figure 1. Statistically significant differences among the information variables: • information needs (mean=3.75) are significantly greater than overall information

provision (mean=3.06) (t=6.048, p<.0001) • internal information needs (3.89) are significantly greater than internal information

provision (3.38) (t=4.06, p<.0001), and • external information needs (3.63) are significantly greater than external information

provision (2.73) (t=6.15, p<.0001).

Interestingly, Table IV shows that the judgmental memory variable has the greatest mean while one would have expected procedural memory to be highest in this group of firms due to its over reliance on rules and procedures and where the employees are routinely evaluated on their compliance with these rules.

The correlations among the independent and the dependent variables are given in Table V.

Table V. Correlation Matrix among the Research Variables. 1 2 3 4 5 6 7 8 9 101 Procedural Memory - 2 Declarative Memory .23 - 3 Judgmental Memory .35* .01 - 4 Software Penetration .48** .02 .29* - 5 Storage Penetration .16 .02 .46** .41** - 6 Network Penetration .48** -.28* .22 .24 .20 - 7 Communication Penetration .11 .17 .13 .50** .38* .30* - 8 Internal Information Needs -.15 .06 .02 -.26* -.21 .23 .11 - 9 Internal Information Provision -.01 -.13 .04 .14 -.08 .29* .37 * .52 *** -

10 External Information Provision .27* .06 .25 .18 .06 .14 .26 * .16 .52 *** -11 External Information Needsa .04 -.20 .16 .03 -.29 .14 -.08 .47 ** .38 * .34* a Adjusted for sign due to correction for skewness by transformation. * ** ***

p <.05 p< .005 p< .0001

Interestingly, from among the three memory variables, only procedural memory and judgmental memory correlate positively with each other (.35, p=.011). There are also significant correlations between the memory variables and the technology variables ranging from .29 to .48. One can notice the .28 (p=.036) negative correlation between declarative memory and network penetration.

Only procedural memory correlates with the dependent information variable measuring external information provision. However, as one would expect, IT correlates with several information variables, either negatively with needs or positively with provision. for example, though not associated with information provision, software penetration is negatively associated with information needs (-.26, p=.046). both network and communication penetrations are associated with internal information provision. The only variable other than procedural memory, correlating

11

with external information provision is communication penetration (.26, p=.044). No independent variable correlates with external information needs.

The data of the correlation matrix raise two issues. First, most independent variables are significantly inter-correlated suggesting multicollinearity might be present. Second, the dependent variables demonstrate several significant inter-correlations. This is restrictive in the sense that running multivariate tests would inflate the possibility of Type I error. Coupled with the likely multicollinearity of the independent variables, the validity of the interpretation of the results could become questionable.

To address the first issue, we used the orthogonal factor scores extracted from the principal components analyses performed initially to extract an orthogonal set of component scores from all the independent variables (Tabachnick and Fidell, 1989). This procedure allowed us to form orthogonal variables which carry the same information as the original variables but which will not correlate, thus reducing multicollinearity.

To avoid the implications of the second issue, we used multiple analysis of variance (MANOVA) to analyze our data. It is inherent in the MANOVA procedure that tests on all univariate and multivariate effects are performed on the dependent variables taken together and separately. To perform these analyses, we split all the independent variables into two groups (low and high) based on their median.

Results

Given the exploratory nature of this study, a series of MANOVAs was conducted investigating main, multivariate and interaction effects between the independent variables and the information dependent variables. No significant results involved the procedural memory and software penetration variables. Furthermore, there were no main, multivariate, or interactive effects on the external information needs dependent variable.

There were, however, main effects and interactive effects involving the other variables. Main and interactive effects involved the storage, network, and communication variables as well as the declarative and judgmental memory variables.

The main effect of storage penetration was on the internal information provision (F=9.30, p=.004). Communication penetration had main effects on both the internal information needs (F=5.05, p=.031) and the external information provision (F=4.62, p=.039) variables.

Four interactive effects were detected. The first was between storage penetration and declarative memory on internal information needs (F=4.15, p=.050, see Figure 2). The second between storage penetration and declarative memory on internal information provision (F=14.70, p=.001, Figure 3). The third between storage penetration and declarative memory on external information provision (F=8.46, p=0.006, Figure 4). And the fourth between network penetration and judgmental memory on external information provision (F=5.20, p=.029, Figure 5).

The results thus clearly suggest that an interaction effect between memory and technological variables on information needs and provision exist.

12

Low HighStorage Penetration

High Declarative Memory

Low Declarative Memory

F=4.15 (p=.050)

Low High

Storage Penetration

High Declarative Memory

Low Declarative Memory

F=14.70 (p=.001)

Figure 2. Interaction effect between declarative memory and storage penetration on the amount of internal information needs.

Figure 3. Interaction effect between Declarative Memory and storage penetration on the amount of internal information provision.

Low HighStorage Penetration

Low Declarative Memory

High Declarative Memory

F=8.46 (p=.006)

Low HighNetwork Penetration

High Judgmental Memory

Low Judgmental Memory

F=5.20 (p=.029)

Figure 4. Interaction effect between Declarative Memory and storage penetration on the amount of external information provision.

Figure 5. Interaction effect between judgmental memory and network penetration on the amount of external information provision.

We also note, from Figure 2, that, between low and high declarative memory, internal information needs are high when there is not enough storage. This result is to be expected since if technical, scientific, and administrative memories exist in the organization, but are not available in electronic form because there is not sufficient storage, the needs for technical, scientific, and administrative information will be augmented by the need to know where to find it. On the other hand, when both declarative memory and storage penetration are high, the need for internal information is low, probably because storage is used to also store this kind of information.

The next two results (Figures 3 and 4) show interaction effects of declarative memory and storage on both internal and external information provision. Both the nature and the form of these interaction effects are unexpected. In fact, one would assume the interaction to be in favor of increasing information provision when both storage and declarative memory are high. The opposite is observed8.

Figure 5 shows that judgmental memory and network penetration interact significantly on the amount of external information provision. Here, too, the same trend appears. In this case, it would seem that external information provision is better when judgmental memory is high and network penetration is low. This result goes counter to our expectations. Judgmental memory

8 The data showed the same pattern of relationships whether the Mahalanobis outliers are included (in the entire sample) or not (in the reduced sample), except that the relationship shown in Figure 4 is no longer significant in the reduced sample.

13

being more a matter of interpretation and experience, the fact that more technology is available does not make it perceived as more useful since it seems that provision gets lower when network penetration goes higher. But this would only explain one part of the results shown in Figure 5, i.e., when both network penetration and judgmental memory are high. We could not find any explanation for the provision being higher when network penetration is low and judgmental memory is low than when they are both high.

Conclusion

The results of this study are somewhat unexpected. This is perhaps due to the lack of empirical studies having tackled the measurement of organizational memory and the inclusion of both IT and organizational memory in an integrated research model. This paper endeavored to overcome this shortcoming by incorporating both IT and memory concepts into the information-processing theory of the firm.

Unfortunately, most of our results do not support the view that IT and memory contribute simultaneously to increase the provision of information in organizations. While assertions found in the literature hypothesized that IT augments the benefits of organizational memory (Ackerman, 1996; Lehner et al., 1998), this study could not confirm all of them. In fact, it would seem as though IT and organizational memory have diverging effects on information provision.

These results are important because they highlight the effects that IT could have on learning in general and on memory in particular. Because organizations expect that IT would augment organizational memory, they would rely overly on it and expand less efforts in using other types of memories. Could this result in a net negative effect on information availability and has such a phenomenon been observed before?

We have already noted that IT-based and other memories differ in significant ways, one being the additional search effort needed to retrieve information and knowledge when they are stored electronically. Declarative memory, for example, has the characteristic of not being committed to a particular use which could become a drawback as was already noted by cognitive psychologists (Singley and Anderson, 1989) who note that “vast amounts of [declarative memory] are potentially relevant in any problem-solving situation, and this leads to serious problems of search” (p.220), a process by which stored information is recognized by its user as relevant to a particular problem or goal (Wijnhoven, 1999). Search presumes knowledge about what to search (the information), where to search it (in computers or elsewhere) and, in the case of IT, how to search it (programs and query routines). With respect to computer-based memories, this is in conformity with Tuomi’s (2000) notion of “reversed knowledge hierarchy” whereby meaningful information emerges only when we have knowledge about it.

If such is the case, then the additional search effort entailed by computerized files and databases would increase the difficulty of having access to the information needed. This would increase the perception that information is less available since its access requires knowledge that may be lacking. Users may find it difficult even to select from the available sources those that are most appropriate given their needs (Lansdale, 1988). Or they may be unsure as to whether knowledge documented in repositories is current or out-of-date (Galunic and Weeks, 1999).

When information is stored in computers, there always needs to be programs to help access it and organizational members need to know how to use these programs. Furthermore, organization members need tools to make sense out of the information found.

Another possible, and somewhat related, explanation could be found in both the psychological expectations of computer users (see Lansdale, 1988) and the Computer Credibility literature (Sheridan et al., 1983), which suggests that the trust that people put into computers may result in poorer performance. For example, when typing a text, people are less careful because they rely on the word-processor’s spell checker; they lose their arithmetic skills because they rely on

14

calculators, and their performance in using word-processor decreases when an intelligent agent, an Advisor, consisting of a context-sensitive function, is turned on (Galletta et al., 2005).

Limitations of the Study No study is perfect and this empirical investigation is no exception. The first limitation is the small sample size which, coupled with the fact that all the work-units studied are affiliated to the same group of companies, precludes us from generalizing our results to other organizations.

The fact that the data was collected via questionnaires constitutes another limitation, especially that one of the advantages of questionnaires studies to tackle large samples has not been realized.

The third limitation is related to the operationalization of the memory variables. Because this was done for the first time and though the measures exhibited acceptable reliability, their content and construct validity are still not established.

Suggestions for Future Research This study opens avenues for future research to test the hypothesis that IT and organizational memory have additive effects. This study failed to show this result. Other measures should perhaps be used, especially in the case of organizational memory.

In Randall et al.’s (2001) study, there was a general preference among workers to ask a local expert rather than interrogate manuals, computer screens, or other forms of knowledge support. If even knowledge workers such as Andersen Consulting consultants more often ask their colleagues for help to locate documents in the firm's repositories (Markus (2001) and if engineers and scientists in search of assistance turn to colleagues five times more often than to a computerized data base (Cross et al., 2001), future research should probably explore what would people choose if they had the choice between IT and other means, and under what conditions.

Is national culture involved? Knowledge and memory have often been the subject of disagreements among researchers for whom knowledge is “a fundamentally cognitive concept” and researchers who consider it as a “socially and culturally” derived phenomenon (Ackermann and Halveson, 1998; Randall et al., 2001). More research is definitely needed studying knowledge management in Arab countries perhaps through cross-comparisons with other cultures.

In this paper, a distinction was made between IT-based memory and other types of memory. Aside from IT and human memory, there might still be a lot of knowledge stored mainly in the form of the organization and in the operation of the organization’s socio-technical system (Lehner et al., 1998; Walsham and Barrett, 2005). Not all kinds of knowledge can be stored electronically; some knowledge could be supported by IT and knowledge could be found in control systems, such as standard procedures, rules and regulations, codes and codifications, descriptions of work processes (Feldman and March, 1981). Future studies might consider incorporating those aspects of organizational memory.

Clearly, the concepts related to organizational memory and their effects are yet to be discovered. Future studies should address the issue of homogeneity of needs or use larger samples if heterogeneity cannot be avoided. The issue of homogeneity has been addressed in several studies (Miner, 1990; Tushman and Anderson, 1986) and is suspected to reduce the value and effect of organizational memory. When organizational memory is heterogeneous, organizational members would find it easier to access the information they need than to have to search it.

It may be premature to conclude that no matter the IT advances in a firm, it stills depends on its overall environment. The fact that this study is one of the first to be performed in Tunisia, a country that Henry (1998) describes as “information-shy” due to its seemingly peculiar approach to information management, may be an incentive to explore the concepts related to information and the way organizational memories are grasped in culturally different backgrounds and environments.

15

[References available upon request from the first author]