A wide range of research has proposed IS success models [6] [8] -[10] . These models postulate their own definitions of IS success and factors that affect the defined IS success, however, a first synthesis of the manifold perspectives on IS success and its underlying antecedents was achieved by DeLone and McLean (1992) [6] , who developed an IS success model. This model has received much attention from IS scholars and can be regarded as one of the most prominent and influential models in IS research [11] . DeLone and McLean (1992) [6] made a major breakthrough. They conducted a comprehensive review of IS success literature published between 1981- 1987 in seven journals in the IS field, and proposed a model of IS success shown in Figure 1.

Since 1992, a number of studies have undertaken empirical investigations of the multidimensional relationships among the measures of IS success such as [12] [13] . In addition to studies that have tested and validated the

DeLone and McLean IS success model (1992), several studies have been published that challenge, critique, or extend the model itself. On balance, these articles have contributed to a better understanding of IS success such as Ballantine et al. (1996) [18] .

Ballantine et al. (1996) [18] assessed the DeLone and McLean model (1992) from a number of different viewpoints and a new model, the 3-D model, has been proposed. First, the independent variable or the critical success factors, factors that may cause success rather than being part of success, have been incorporated to the 3-D model. Second, the causal relationship that exist between the success dimensions in DeLone and McLean model (1992) has been criticized arguing that, success in one dimension does not lead directly to success at the next dimension. If this were so, then one need only be successful at system quality to ensure the success of the whole IS. Finally, Ballantine et al. (1996) [18] mentioned that DeLone and McLean (1992) [6] provide a unidirectional model which moves from system and information quality to organizational impact of IS, the impact upon organizational learning is not discussed. The implication in the DeLone and McLean model (1992) is that success is necessary at each stage otherwise the next stage will not be successful. It is not sufficient to move only in one direction. Unless individuals and the organization can learn from experience and develop better systems and recognize better information quality then it is unlikely that the measurement of IS outcome will serve any useful purpose. Thus, Ballantine et al. (1996) [8] depict the importance of organization learning through time and experience in their model through the learning feedback loop as shown in Figure 2. The 3-D model separates success into three fundamental levels: the technical development level, the deployment to the user, and the delivery of business benefits. Hence it is termed the 3-D model as shown in Figure 2. In the 3-D model, filters act between the levels of IS success and contains influences which inhabit or encourage the adoption of the IS at the next higher level. Success at each level is influenced by a number of different independent variables and the outcomes of each level (development, deployment, and delivery) are the closest equivalent to dependent variable (outcome measures of IS success).

Based on suggestions offered by researchers, and criticisms directed to the DeLone and McLean original model, DeLone and McLean (2003) [9] proposed an updated version of their IS success model (1992) and evaluated its usefulness in light of the dramatic changes in IS practice, especially the advent and explosive growth of e- commerce. The updated model is shown in Figure 3. The primary differences between the original and updated models include:

1) The addition of service quality to reflect the importance of service and support in successful e-commerce systems;

2) The addition of intention to use to measure user attitude;

3) The collapsing of individual impact and organizational impact into the net benefits construct.

The categories of the updated taxonomy were system, information, and service quality, intention to use, use, user satisfaction, and net benefits. DeLone and McLean models (1992, 2003) could serve as a basis for the selection of appropriate IS measures. Researchers had to choose several appropriate success measures based on the objectives and the phenomena under investigation, as well as consider possible relationships among the success dimensions when constructing the research model [14] .

While the updated DeLone and McLean model (2003) is a comprehensive IS success model, it suffers from certain difficulties. First, the Net Benefit measure in the model is conceptually too broad to define. As [15] suggests, “The new net benefits construct immediately raises three issues that must be addressed: what qualifies as a benefit? for whom? and at what level of analysis” (p. 32). Thus, when using the updated DeLone and McLean model, researchers need to clearly and carefully define the stakeholders and the context in which Net Benefits are to be measured.

Second, as DeLone and McLean (1992) [6] stated “The selection of IS success measures should also consider the contingency variable such as the independent variable being researched: the organizational strategy, structure, size, and environment of the organization being studied; the technology being employed; and the task and individual characteristics of the system under investigation” (p.88). Thus, DeLone and McLean (1992) [6] recognized the limited perspective of their model. They identify only the dependent variables (outcome measures) of IS success.

While the 3-D IS success model represents a holistic view of the concept of IS success. This model did not focus on the outcome measures of IS as DeLone and McLean (1992, 2003) did. Although Ballantine et al. (1996) [8] incorporated the critical success factors (independent variables) to their success model, the outcome measures of IS success are unclear and misspecified.

Based on the above mentioned literature, this study proposes a new GIS success model by extending and respecifying both the 3-D model of IS success and the updated DeLone and McLean model (2003) in the context of GIS.

This level extends and respecifies the 3-D model of IS success developed by Ballantine et al. (1996) [8] to GIS context by identifying the CSFs in each stage of GIS diffusion process (instead of development, deployment, and delivery levels in the 3-D model) that influences the success of GIS project adoption at each stage. The concept of CSFs was popularized in the context of IS and project management (PM) by [20] as factors affecting the success of activities and projects. CSF is defined by [21] as “the limited number of areas in which satisfactory results will ensure successful competitive performance for the individual, department, or organization. CSFs are the few key areas where ‘things must go right’ for the business to flourish and for the managers goal to be attained” (p. 385). CSFs make it easier for managers to prioritize vital aspects of a project [22] .

According to [23] diffusion is “the fundamental process that is responsible for the transfer of innovations from the workshops of their inventors to becoming a daily part of the lives of a large section of society”. GIS project diffusion can be classified into three main stages [24] - [27] : (1) initiation (pre-implementation), (2) acquisition (implementation), and (3) incorporation (post-implementation). Detailed descriptions of these stages and the steps carried out in each one are reported elsewhere (see, e.g., [28] [29] , for an excellent overview). The main objective of the first level of the proposed GIS success model is to identify the CSFs in each stage of the GIS project diffusion that influence the success of GIS project adoption at each stage as shown in Table 1. Some CSFs may work at more than one stage. The CSFs identified in the three stages of GIS diffusion process were extracted from:

1) GIS success research cited in the literature which is mostly based on case studies or observations of GIS projects and practices, such as [30] - [32] ;

2) GIS failure research which is based on lessons learned from certain types of GIS projects, but they are mostly similar enough to be generalized, such as [33] - [35] ;

3) Researches about GIS implementation that mentioned CSFs briefly such as [36] - [38] .

The second level of the proposed model extends and respecifies the updated DeLone and McLean model (2003) to the GIS context via the following steps:

1) Replacing service quality dimension (that reflect the importance of service and support in successful e-com- merce systems) with user quality. As GIS is not solely technical in nature, adding user quality dimension as a part of the process of producing geographic information is very important. GIS user quality is considered an important human factor in a successful GIS as mentioned in many GIS researches [62] ;

2) According to [15] , researchers need to clearly and carefully define the stakeholders and the context in which Net Benefits are to be measured. Therefore, we split the net benefits dimension to net benefits to individuals, organization, and society;

3) DeLone and McLean IS success models (1992, 2003) are built upon the taxonomy developed by both Shannon and Weaver (1949) [64] and Mason (1978) [65] , which considered that a message in a communication system can be measured at different levels including: the production level, the product level, and the influence level. The production level is the accuracy and efficiency of the system which produces the information. The product level is the success of the information in conveying the intended meaning. The influence level is the effect of the information on the receiver. The influence level is presented as a hierarchy of events taking place at the receiving end of an information system; these events are receipt of the information; influence of the information on the receiver; and the influence of the information on the performance of the system. The concept of levels of output from communication theory demonstrates the serial (process) nature of information (i.e., a form of communication). The IS creates information which is communicated to the recipient who is then influenced (or not) by the information. In this sense, information flows through a series of stages from its production through its use or consumption to its influence on individual and/or organizational performance [6] . Therefore, to reflect the interdependent and the process nature of the success dimensions that was suggested by DeLone and Mclean in their models (1992, 2003), we believe that system quality (which measures the success of the production level) and information quality (which measures the success of the product level) should not happen in parallel as DeLone and McLean (1992, 2003) did in their models. Instead, information quality should come after the system quality dimension to reflect the interdependent and the process nature of the success dimensions;

4) To avoid model complexity and to reflect the cross-sectional nature of this study, the feedback links from net benefits to both use and user satisfaction in the updated DeLone and McLean model (2003) were excluded.

The success dimensions of the proposed GIS success model are system quality, user quality, information quality, information use, user satisfaction, net benefits to individuals, net benefits to organization, and net benefits to society as shown in Figure 4. The arrows between success dimensions represent temporal (process) and causal relationships. The following section will discuss the GIS success dimensions.

The second level of the proposed model is concerned with measuring the GIS success after incorporating and adapting the GIS into the organization's operations. The GIS success should be measured after a wide spread of use to allow the members of the organization to arrive at informed opinions about the success of their GIS [29] . Hence it is termed GIS post-implementation success.

In this paper, according to previous researches on GIS, measures of the proposed model have been determined (see Table 2).

Ÿ System quality: system quality dimension measures the success of the technical aspects of GIS. System quality has been represented in many GIS studies by functionality, response time, system reliability, user friendless, error recovery, database content.

Ÿ User quality: user quality dimension represents the quality of GIS users in terms of spatial abilities and self efficacy. In IS field, Bonner (1995) [66] revised the DeLone and McLean model and introduced user quality in terms of knowledge skills and abilities. His recognition of the people element was a welcome addition to the model. Also, the recognition of the importance of human factor in evaluating GIS performance was first initiated by [67] , who stated that people, not the computerized equipment, make a GIS success or fail.

Ÿ Information quality: information quality dimension is the quality of information provided to the organization using GIS, in the form of maps, tables, charts, and reports. The information quality dimension measured by accuracy, completeness, ease of interpretation, relevancy, reliability, timeliness, and clarity.

Ÿ Information use: information use is a broad construct that is frequently used in measuring the utilization of IS. Information use dimension measures to what extent the GIS output is being used in the decision making process. Clapp et al. (1989) [18] mentioned that, the system can provide the capability to obtain all desired information, but for some reasons, the information is not used in the decision process whether private or public. In this case, the GIS will fail due to lack of utilization. Use can be based on objective measures such, number of functions used, frequency of access, and amount of connecting time [68] . Questions about who uses the system, levels of use, motivations for and voluntariness of use, and the purpose and nature of system use are also relevant [67] .

Ÿ User satisfaction: this dimension measures GIS user’s level of satisfaction with the system. User satisfaction was traditionally employed as the most common measure of IS success. The most widely used user satisfaction instruments are End User Computing Support (EUCS) [69] instrument and User Information Satisfaction (UIS) [70] . Both the EUCS and the UIS instruments contain items related to system quality, information quality, rather than only measuring overall user satisfaction with the system. Because of this, some researchers have chosen to parse out the various quality dimensions from these instruments, and use a single item to measure overall satisfaction with an IS [71] .

Ÿ Net benefits to individuals: this dimension summarizes benefits that can be gained by users when using GIS such as enhanced decision making, time saving, increase the understanding and awareness of problems [72] .

Ÿ Net benefits to organization: this dimension summarizes the benefits that organization derives from using GIS, which refers to efficiency and effectiveness criteria. Efficiency is the degree to how GIS operates with minimum waste, duplication, and expenditure of resources, and can be expressed as cost savings, cost avoidance, or productivity gains. Efficiency may also result in the generation of revenue. Effectiveness involves generating a product of better quality or accomplishing an intended purpose [67] .

Ÿ Net benefits to society: based on the study of “The impact of GIS technology” conducted by [67] , “societal impact” dimension has been proposed as a further variable to the lists of six DeLone and McLean’s IS success dimensions. Societal impact is important to be considered in the evaluation of GIS success because the ultimate goal of all technologies introduced in the public sector agencies is to benefit society. Many researchers reported the benefits of GIS on broad societal objectives such as, citizen-public sector interactions, individual integrity, economic benefits, distribution of wealth and fulfillment of human aspirations, and equity.

The first level of the proposed GIS success model acts like a guide or a reference for GIS project managers to concentrate on the most critical success factors of GIS project diffusion. Although, establishing the CSFs of GIS project does not implicate that the whole project will automatically succeed, but it would be erroneously to neglect one of these CSFs.

The second level of the proposed GIS success model is a multidimensional construct, and the dimensions are interrelated. GIS are first implemented and incorporated within the organization and exhibit various degrees of system, and user quality. System quality and user quality affect the quality of the produced information. Managers/decision makers experienced the quality of information by using it for their works. Users and managers/de- cision makers are either satisfied or not satisfied with using the GIS. Finally, the use of information by managers/decision makers and the satisfaction of GIS users trigger influence on net benefits to individuals, organization and society.

The second level of the proposed model suggests that there can be positive influence between the GIS success dimension. Thus, we propose the following 11 hypotheses:

H1. System quality will positively affect Information quality;

H2. User quality will positively affect Information quality;

H3a. Information quality will positively affect user satisfaction;

H3b. Information quality will positively affect information use;

H4. Information use will positively affect User satisfaction;

H5a. Information use will positively affect net benefit to individuals;

H5b. Information use will positively affect net benefit to organization;

H5c. Information use will positively affect net benefit to society;

H6a. User satisfaction will positively affect net benefit to individuals;

H6b. User satisfaction will positively affect net benefit to organization;

H6c. User satisfaction will positively affect net benefit to society.

To ensure the content validity of the scales used in this study, We used the measurement items that were operationalized and tested in previous empirical GIS/IS studies and were found to have demonstrated good psychometric properties. The measuring items for each success dimension are summarized in Table 2.

The data used to test the model were obtained from a sample of experienced GIS users and managers. This study developed a questionnaire (see appendix A) using a five-point Likert scale (1 - 5) ranging from “strongly disagree” to “strongly agree.” The questionnaire was sent to 350 GIS users and managers in different GIS organizations in Egypt and abroad to answer the questions by assessing their GIS. For each question, respondents were asked to circle the response which best described their level of agreement. In total, 252 samples were received with an effective ratio of 72%. Detailed descriptive statistics relating to the respondents’ characteristics are shown in Table 3.

Reliability refers to the consistency or stability of the questionnaire results. Fewer errors lead to a higher level of reliability. In other words, a better reliability measurement will result from the consistency and stability of results. The present study measured the questionnaire reliability and the consistency of the items using Cronbach’s alpha (SPSS Version 20). Many scholars have suggested that a Cronbach’s alpha coefficient exceeding the 0.7 threshold indicates a high level of consistency among the aspects; a Cronbach’s alpha coefficient exceeding 0.9 indicates a much higher level of consistency among the aspects (e.g., [79] - [83] ). The Cronbach’s alpha coefficients of the eight construct are greater than or equal the recommended value of 0.8 (see Table 4).

Structural Equation Modeling technique was used to assess the model fit and show empirical findings and hypotheses results using LISREL 8.8. Seven common model-fit measures were used to assess the model's overall goodness of fit: the ratio of X² to degrees-of-freedom (df), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), normalized fit index (NFI), comparative fit index (CFI), root mean square residual (RMSR), and root mean square error of approximation (RMSEA). As shown in Table 5, all the model-fit indices exceeded their respective common acceptance levels suggested by previous research, thus demonstrating that the proposed model exhibited a fairly good fit with the data collected. Thus, we could proceed to examine the path coefficient of the structural model.

Properties of the causal path, including the path coefficient, t-values, and variance explained for each equation in the hypothesized model, are presented in Figure 5.

System quality and user quality had significant positive influences on information quality. Thus, H1 and H2 were supported. The influence of information quality on both user satisfaction and information use were also significant. H3a and H3b were supported. The influence of information use on user satisfaction was not significant. Thus, H4 was rejected. User satisfaction has no significant impact on net benefits to (individuals, organization, and society). H6a, H6b, and H6c were rejected. Information use has a significant positive influence on both net benefits to individuals and net benefits to society, but has no significant effect on net benefits to organization. H5a and H5b were supported, while H5c was rejected. The direct, indirect, and total effects of system quality, user quality, information quality, information use and user satisfaction on net benefits to (individuals, organization, and society) are summarized in Table 6.

With regard to the constructs explained in the variance (R²), 89% of the variance in information quality was explained by system quality and user quality, while 20% of the variance in information use was explained by information quality. 87% of the variance in user satisfaction was explained by information quality and information use. The variance explained by information use and user satisfaction on net benefit to individuals is 34%, net benefit to organization is 8%, and net benefit to society is 13%.