Using MCDA to select countermeasures against fake news

Journal of Information, Communication & Ethics in Society Using MCDA To Select Countermeasures Against Fake News Journal: Journal of Information, Communication & Ethics in Society Manuscript ID JICES-07-2024-0089.R1 Manuscript Type: Journal Paper Keywords: MCDA, M-Macbeth, Risk Management, Fake News, Disinformation Journal of Information, Communication & Ethics in Society Journal of Information, Communication & Ethics in Society Using MCDA To Select Countermeasures Against Fake News Abstract The advent of fake news poses a significant threat to communities, organisations and individuals, contributing to the erosion of public trust in institutions and democracy. This is aggravated should we consider the multiplicity of fake news and, thus, the multitude of risk and their impact on society. This research proposes tackling fake news as a digital risk by applying a Multi-Criteria Decision Analysis (MCDA) to select the appropriate countermeasures for law enforcement agencies (LEAs) to tackle high stakes crime associated with fake news. Here, we demonstrate the applicability of MCDA through a systematic approach using M-Macbeth to present alternatives for mitigating high-impact instances of crime in a community and measure the attractiveness of each alternative. Results indicate that to mitigate risk effectively, prioritising risk using adequate strategies and appropriate courses of action is crucial. Nevertheless, the contributions of this research work allowed us to comprehend the best alternative to mitigate the risk of fake news and provide a realistic approach to support LEAs in decision analysis. Key Words: MCDA, M-Macbeth, Risk Management, Fake News, Disinformation Page 1 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society 1. Introduction Fake news (FN) is not just a term but a pressing issue that demands our immediate attention. It refers to the widespread dissemination of false or misleading information, intentionally or unintentionally. This urgent phenomenon poses a significant risk as it can manipulate public perception and influence societal dynamics. FN is closely related to two key terms— disinformation and misinformation—which are often used interchangeably but represent distinct concepts that are important to differentiate [1]. Disinformation refers to the deliberate creation and spread of falsehoods, usually driven by motives such as political, financial, or ideological gain [2]. It is a calculated effort to deceive and mislead, as recognised by institutions such as the European Union and the British Parliament, which acknowledge disinformation as a fundamental problem in modern society [3], [4], [5]. On the other hand, misinformation occurs when individuals unknowingly share inaccurate information, believing it to be true. While disinformation is intentional, misinformation is accidental. However, both contribute to the erosion of public trust, social division, and significant risks, especially in the digital age where fake news rapidly circulates on social media platforms [2]. The issue of FNs is not modern—it has existed throughout history in various forms, from the propaganda tactics of ancient Rome to the press manipulation during World War II [6], [7]. More recently, the ongoing conflict between Ukraine and Russia has demonstrated how false information plays a pivotal role in the strategies of modern warfare. This historical context underscores the gravity of the issue. The difference today lies in the advancement of technology, particularly the rise of social media, which acts as a catalyst, allowing FNs to spread swiftly and more widely, exacerbating its threat to organisations, individuals, and society [7]. This paper explores the intersection of fake news as a threat and established risk management principles, particularly those outlined in the ISO 31XXX family of standards [8]. By examining risk management strategies, the paper proposes a method for mitigating the digital risks of FNs using Multi-Criteria Decision Analysis (MCDA). This approach enables decision-makers to rank potential countermeasures and identify effective strategies to combat the high-impact crime of disinformation within a community. Page 2 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Traditionally, risk management strategies have been applied to address financial, operational, or strategic uncertainties, including those arising from the spread of FNs. A fundamental principle of risk management is that anything of value is worth protecting [8]. While this principle is often applied in organisational contexts, it is equally essential to safeguard our communities from the multifaceted risks of FNs. Effective strategies must be adopted to prevent the mismanagement of high-impact threats. Law Enforcement Agencies (LEAs) across the European Union are not just tasked with investigating crimes related to FNs but with proactively managing and preventing them. They must act quickly when faced with high-stakes situations. This paper strongly advocates for a proactive approach to managing disinformation risks, which have been identified as both severe and highly likely to occur in specific communities. By providing LEAs with a structured method for responding to such risks, this work offers a framework to support more effective decisionmaking and mitigate the negative consequences of fake news. The structure of this paper is as follows: The next section presents a research background on risk management, MCDA, and the intersection of these two subjects. A detailed explanation of the design science research methodology follows this. The fourth section introduces the author's proposal, followed by a demonstration and evaluation of the findings. Finally, the paper concludes with a summary of the key insights, a discussion of the research limitations, and suggestions for future work. 2. Research Background This section delves into Risk Management, Fake News, and MCDA. This segment encapsulates a comprehensive exploration of these key topics, providing the foundation for understanding our research's intricacies and interconnections. 2.1 Risk Management In today's increasingly uncertain world, it is essential to adopt effective strategies that help organisations and communities anticipate potential risks and mitigate their impact through appropriate countermeasures. One of the most widely recognised frameworks for managing risk is the ISO 31000 standard, which provides guidelines and principles to address risks systematically and effectively [8]. The standard offers a structured framework that can be adopted by organisations, governments, and other entities to implement policies, procedures, Page 3 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society and practices designed to manage risks comprehensively. These activities include communicating and consulting with stakeholders, setting the context, and continuously assessing, treating, monitoring, and reporting on risks [8]. The ISO 31000 risk management process consists of several vital steps. First, it emphasises the importance of communication and consulting to ensure that all relevant parties are informed and involved in risk management decisions. Next, the scope, context, and criteria for risk assessment are established to define the boundaries and objectives of the risk management process. Following this, a thorough risk assessment is conducted, which includes identifying, analysing, and evaluating risks. Once risks are assessed, the next step is risk treatment, where appropriate measures are taken to mitigate or manage the risks. The framework also highlights the importance of monitoring and reviewing the effectiveness of risk management actions and recording and reporting to ensure transparency and accountability throughout the process [8]. In addition to the ISO 31000 guidelines, alternative approaches to risk management are being proposed to meet the evolving challenges of the modern world. For example, Chapter 5 of Corporate Risk Management advocates for rethinking traditional risk management practices, arguing that risk management should be about avoiding losses and creating value for businesses [9]. This perspective shifts the focus from a defensive approach to one that sees risk management as an opportunity for strategic growth and value creation. As we delve deeper into the nature of risk, it is crucial to understand that the digital landscape presents its own set of challenges. In the next section, the paper will explore the digital risks posed by FNs, summarising their nature and impact better to contextualise the threat within modern risk management strategies. 2.2 Fake News The term Fake News (FN) is often associated with a wide range of terminology, making it somewhat challenging to differentiate between similar concepts. Terms such as disinformation, misinformation, false information, fabricated information, and partisan information are frequently used in connection with FN, each with its nuanced meaning [10]. A standard definition of FN is the intentional or unintentional spread of false or misleading information in Page 4 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society the public sphere, with the potential to cause harm to organisations, individuals, public institutions, and the media [7]. FN spreads effectively in the modern era, largely due to the proliferation of online platforms, with social media being a key player in its dissemination [1]. State or private actors often exploit these platforms to spread FNs, employing sophisticated techniques to propagate falsehoods. These methods can include using automated bots that rapidly disseminate false information across networks or infiltrating real social media accounts to lend credibility to FNs [11]. While FN is often associated with social media, it is important to note that its reach extends beyond digital platforms. Fake news can also be found in traditional media forms, such as newspapers, newscasts, and other periodicals, demonstrating that false information can take many forms and use various channels for dissemination [12]. Understanding the intent behind the spread of FN is crucial in distinguishing between disinformation and misinformation. If the spread of false information is a deliberate act intended to mislead, it is classified as disinformation. On the other hand, if the spread results from an error or mistake, it is considered misinformation [2]. Misinformation can also include incomplete information that leads to misunderstandings [7]. Literature suggests that private interests, often seeking political or financial gain, target vulnerable individuals to act as unwitting participants in the further spread of FN [13]. Malicious actors typically employ three common strategies when targeting organisations with fake news: 1. They spread false information to achieve financial gain or to discredit and manipulate political discourse. 2. They often label their distorted version of reality as the only factual source of information, deliberately misleading audiences. 3. In a more insidious tactic, they accuse legitimate news outlets of spreading fake news [7]. Understanding the behaviour and motivations of these perpetrators is crucial for assessing the risks associated with FN and developing strategies to counter its harmful effects. Page 5 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society 2.3 Multi-criteria Decision Analysis Multi-Criteria Decision Analysis (MCDA), also known as Multi-Criteria Decision Making (MCDM), is a decision-support technique designed to assist in complex decision-making processes that involve multiple criteria. These criteria can often be conflicting, making decisionmaking challenging. MCDA offers a structured and systematic approach to simultaneously evaluate various factors, improving the overall quality of decisions [14], [15], [16]. It is widely used in healthcare, education, and public policy, where decisions must account for diverse and sometimes competing interests. The strength of MCDA lies in its ability to evaluate and rank alternative options based on the importance of different criteria and how well the alternatives perform relative to these criteria, all while considering multiple stakeholder perspectives [17]. Two critical steps are necessary to make effective decisions using MCDA. First, the problem must be clearly and comprehensively defined, along with its limitations or constraints. Second, it is essential to establish the roles of the key participants in the decision-making process. The Decision-Maker (DM), whether a CEO, doctor, manager, judge, or consumer, is ultimately responsible for making the final decision. The Decision Analyst (DA) systematically evaluates alternatives and provides a structured framework to rank these options. This framework is crucial for guiding the DM toward informed choices. The role of the DA is to build a detailed decision model that accurately reflects the problem at hand. This model ensures that the evaluation process is rigorous and considers all relevant factors. MCDA is an essential tool in this process, offering a structured approach for assessing and prioritising options in situations where multiple criteria are involved, and consensus is required. It is important to note that MCDA does not dictate decisions; rather, it supports the decisionmaking process by providing a clear and logical structure for evaluating options. Every decision, whether consciously or not, involves balancing multiple factors. MCDA helps make this balancing process explicit and more manageable, ensuring the objectivity and fairness of the decision-making process [14]. A common approach within MCDA is using a hierarchical value tree, which structures the decision problem by breaking it down into smaller, more manageable components. This value tree helps assess the value function, representing the DM's preferences. The preferences in the decision process must follow the principle of preferential independence, ensuring consistency Page 6 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society in evaluating options. Another critical aspect of MCDA is elicitation procedures, which involve gathering insights directly from decision-makers [14]. These procedures ensure that the decision process is collaborative, aligning with the real-world preferences and needs of the stakeholders involved. In summary, MCDA provides a robust framework for addressing complex decision-making scenarios by evaluating multiple criteria, considering stakeholder perspectives, and supporting decision-makers in making informed, well-structured choices. 3. Research Methodology This section outlines the methodology employed, emphasising the utilisation of Design Science Research (DSR). 3.1. Design Science Research The proposed methodology – Design Science Research – aims to create innovative and practical solutions to solve complex problems. This objective is achieved through the design of an artefact and its consequent demonstration and evaluation. This research follows the guidelines for correctly implementing DSR in Information Systems [18]. These guidelines encompass the following necessary steps for the correct implementation of the methodology: 1Identification of the Problem and Motivation - define the research problem, highlighting the importance of finding a solution. 2Definition of Objectives - Elaborate research objectives, taking into consideration their feasibility. Objectives can be of quantitative or qualitative nature. 3Design and Development - Creation of the artefact, its desired functions, and architecture. 4Demonstration - in an appropriate activity that solves one or more instances. 5Evaluation - testing the artefact's effectiveness by measuring and comparing the achieved and proposed objectives. 6Communication - essential conveying all the necessary information about the problem underlying its significance, explaining how the developed artefact aids in solving the proposed problem and underlying its effectiveness to relevant audiences. It is important to understand that these steps suggest a problemand objective-oriented iterative process to design and develop a solution following the above nominal process sequence [18]. Page 7 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society 4. Research Design This section includes a short explanation of the research objectives, how MACBETH helps achieve results, and a final explanation of the research proposal. 4.1 Research Objectives This research lies in a twofold approach to tackling the intricate problem of fake news. Firstly, we seek to uncover and define the crucial criteria that guide the selection of countermeasures against the spread of FN. Grasping these criteria is paramount for crafting effective alternatives to combat FN-related crime. One such strategy is developing an MCDA-based framework for strategic decision-making when dealing with high-stakes FN. Another strategy is to provide recommendations for effectively tackling disinformation-related crime based on an MCDA analysis, thus laying the foundation for more focused and efficient countermeasures. 4.2 Utilising MACBETH MACBETH, which stands for Measuring Attractiveness by a Categorical Based Evaluation Technique, is a decision analysis methodology commonly used in MCDA that considers multiple criteria or factors for evaluating and comparing different alternatives[19]. This method is designed to aid stakeholders in making decisions in situations where there are conflicting criteria preferences. Its objective is to provide a structured and systematic approach for evaluating and ranking alternatives using a categorical ranking system to assess the relative importance of the criteria and the performance of alternatives. The assignment of the DM qualitative descriptors (e.g., very weak, weak, moderate, strong, very strong) aids in expressing their preferences [17]. Another important aspect of this method is that it helps with modelling the preferences and values of DM by comparing criteria pairwise. A comparison of criteria allows for a derivation in a numerical scale for each criterion using a mathematical operation to convert qualitative preferences into numerical values, therefore allowing for the quantification of subjective judgments consistently. This methodology incorporates consistency checks, ensuring the DM preferences are logical and contradictions-free. This way allows for the flagging of inconsistencies and their consequent review. Once the preferences for the criteria and alternatives are established, they are aggregated, thus generating an overall ranking or score for each alternative[17], [19]. In summary, MACBETH is a Page 8 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 4. Value Function for the Criteria 1 (Impact) Figure 5. Differences in attractiveness between the performance of the options on the Criteria 2 (Context) Page 15 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 6. Value Function (Numerical scale related to performance levels) Figure 7. Differences in attractiveness between the performance of the options on the criteria 3 (Type of Event) Page 16 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 8. Value Function (Numerical scale related to performance levels) Figure 9. Differences in attractiveness between the performance of the options on the Criteria 4 (Spread of the Fake News) Page 17 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 10. Value Function (Numerical scale related to performance levels) Figure 11. Differences in attractiveness between the performance of the options on the Criteria 5 (Probability) Page 18 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 12. Value Function for the Criteria 5 (Probability) It is important to note that the M-MacBeth software tool offers a scale of scores, with the lower one receiving a score of 0 and the higher one receiving a score of 100, following the references earlier specified in the attributes of each criterion. This scale displayed by the software can be changed should the DM find one disagreeable criterion, per his opinion. Except for the Impact value function, the DM agreed with all the other four value functions (Context, Type of Events, Spread of the Fake News and Probability). For the Impact criteria, the DM requested that the performance of the option "Direct Communication with Agent(s) of Disinformation (DCA)" be modified from 50 to 65. Changes that resulted in the scale are shown and referenced in Figures 12 and 13. Figure 13. Differences in attractiveness between the performance of the options Page 19 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 14. Value Function for Criteria 1 (Impact) with new DCAD of 65% 5.3.2 Weighting Coefficients After the DM provided his judgment about the difference in attractiveness between the performance profile of the options in each criterion, the DM was then asked to rank the difference in overall attractiveness between the criteria. Simple weighting was used after introducing the judgments in the decision-maker's position, in descending order (from left to right) of attractiveness. With reference levels for each criterion, it was possible to compare the level of attractiveness between the swings of each criterion. The DM approved the weights calculated for each criterion, with no changes having been made. Figure 15. Matrix of attractiveness judgments regarding the difference in global criteria. Page 20 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 16. The scale of the weights obtained for each criterion. 5.4 Application of the Additive Model Using the additive value model in the M-MacBeth software, we could evaluate the attractiveness of each alternative after developing our model. As a result, the program uses the additive model to determine the overall scores for each option. 5.4.1 Computation of the Global Attractiveness of Each Option The scores of each option in each criterion and the overall score are determined using the M-MacBeth software. The attractiveness values of the options are then determined using our additive model's value functions, criteria weights, and performance profiles. Page 21 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society Figure 17. Option scores according to the Additive Model Figure 18. Option Overall Thermometer The BSD (Ban the Source of Disinformation) option is the most appealing one, according to the results of Figures 17 and 18 above. Page 22 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society 6. Evaluation To evaluate the decision model, the authors used a sensitive and robustness analysis. The sensitive analysis aims to understand the impact of uncertainties or fluctuations in the input variables of our model of decision. On the other hand, a robustness analysis refers to an examination of the stability and reliability of the decision-making model under various conditions and uncertainties; the goal is to assess how well the chosen options perform when faced with changes in criteria weights, variations in the input data or other sources of uncertainty [20]. 6.1 Sensitive Analysis Though the DM established the weights allocated to each option, he has the right to change them if unsure how to evaluate the criteria weights. As a result, it is possible to analyse the degree to which the model's recommendations vary when the weight of criteria is modified (while keeping the proportionality relationship between the other weights) by performing a sensitivity analysis on the weight of criteria. By analysing the sensitivity to each of the criteria's weights, we may evaluate the situations in which the most attractive option changes. To visualise the results of the sensitive analysis for each criterion, please consult the appendix of this article. A sensitivity analysis demonstrated that a relatively insignificant change of 22.27% in a particular parameter substantially alters the relative attractiveness of alternative options, particularly emphasising DCAD's competitive edge over BSD. This narrow range of variation highlights the delicate balance between these two choices, suggesting that a slight adjustment within this range could elevate DCAD to parity with or even surpass BSD in terms of attractiveness. 6.2 Robustness Analysis There is always uncertainty and imprecision in decision-making practices and modelling to select options, making it difficult to determine whether the most attractive options are robust. Robustness analysis should determine if an option remains the most attractive despite changes in option performance and criteria weight. It is a crucial aspect of MCDA, evaluating the sensitivity of outcomes and uncertainties in input data, model assumptions, and DM preferences. By identifying influential factors, robustness analysis enhances decision-making confidence and validates MCDA results, ensuring their stability across scenarios [21]. In other words, the global attractiveness rankings for each option differ if the option performs poorly on the criteria or changes the criteria weights. Page 23 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society To perform this analysis, the authors use M-Macbeth's robustness analysis feature. This feature introduced uncertainty in the performance and criteria weighting options in the local and global information sections. First, as shown in Figure 41, the BSD option is recommended and robust without introducing any uncertainties into our model. Figure 19. Robustness Analysis In the figure above, additive dominance is present. Of this additive dominance, we see that the BSD option additively dominates all the other options. Option EP is dominated by all other options (additively dominated by BSD and IMST) and dominated by DCAD and BCP in all criteria. In consultation with the DM, the following degrees of uncertainty were considered in the analysis: Uncertainty of 10% for criteria impact, Uncertainty of 5% for criteria probability. Page 24 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society [17] C. A. Bana e Costa and M. P. Chagas, ‘A career choice problem: An example of how to use MACBETH to build a quantitative value model based on qualitative value judgments’, Eur J Oper Res, vol. 153, no. 2, pp. 323–331, Mar. 2004, doi: 10.1016/S0377-2217(03)00155-3. [18] A. Hevner and S. Chatterjee, Design Science Research in Information Systems. Boston: Springer, 2010. doi: 10.1007/978-1-4419-5653-8_2. [19] C. A. Bana e Costa and J. C. Vansnick, ‘MACBETH — An interactive path towards the construction of cardinal value functions’, International Transactions in Operational Research, vol. 1, no. 4, pp. 489–500, Oct. 1994, doi: 10.1016/0969-6016(94)90010-8. [20] I. Dimitrakopoulos and K. Karamanis, ‘Decision Making Using Multicriteria Analysis: A Case Study of Decision Modeling Career in Education’, Case Studies in Business and Management, vol. 4, no. 2, p. 24, Jul. 2017, doi: 10.5296/CSBM.V4I2.11355. [21] C.-L. Hwang and A. S. Md. Masud, ‘Multiple Objective Decision Making — Methods and Applications’, vol. 164, 1979, doi: 10.1007/978-3-642-45511-7. Page 31 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society i Appendix INDEX I. SENSITIVE ANALYSIS FOR THE CRITERIA 1 – IMPACT......................................................................II II. SENSITIVE ANALYSIS FOR THE CRITERIA 2 – CONTEXT .................................................................IV III. SENSITIVE ANALYSIS FOR THE CRITERIA 3 – TYPE OF EVENT\ ................................................VI IV. SENSITIVE ANALYSIS FOR THE CRITERIA 4 – SPREAD OF THE FAKE NEWS ........................VIII V. SENSITIVE ANALYSIS FOR THE CRITERIA 5 – PROBABILITY...........................................................X Page 32 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society ii I. Sensitive Analysis for the Criteria 1 – Impact Figures 2227 display the sensitivity analysis results concerning the weight of the impact criteria (Imp). Figure 22. Sensitivity analysis on the weight of the criteria impact Figure 23: Sensitivity analysis of the weight of the criteria impact for the BCP and BSD options Page 33 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society iii Figure 24. Sensitivity analysis to the weight of the criteria impact for the BCP and DCAD options Figure 25. Sensitivity analysis to the weight of the criteria impact for the BSD and DCAD options Figure 26. Sensitivity analysis to the weight of the criteria impact for the EP and IMST options Figure 27. Sensitivity analysis to the weight of the criteria impact for the BCP and IMST options Page 34 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society iv The sensitivity analysis of the weight of criteria 1 (impact) from figures 22 to 27 revealed: Weight of impact criteria < (below) 16.9% IMST is more attractive overall than BCP, Weight of impact criteria = 16.9%: BCP and IMST are equally attractive, Weight of impact criteria > (above) 16.9%: BCP is more attractive overall than IMST. Weight of impact criteria < (below) 58.5% DCAD is more attractive overall than BCP, Weight of impact criteria = 58.5%: DCAD and BCP are equally attractive, Weight of impact criteria > (above) 58.5%: BCP is more attractive overall than DCAD. Weight of impact criteria < (below) 65.6% BSD is more attractive overall than BCP, Weight of impact criteria = 65.6%: BCP and BSD are equally attractive, Weight of impact criteria > (above) 65.6%: BCP is more attractive overall than BSD. Weight of impact criteria < (below) 72.6% BSD is more attractive overall than DCAD, Weight of impact criteria = 72.6%: BSD and DCAD are equally attractive, Weight of impact criteria > (above) 72.6%: DCAD is more attractive overall than BSD. Weight of impact criteria < (below) 95.3% IMST is more attractive overall than EP, Weight of impact criteria = 95.3%: IMST and EP are equally attractive, Weight of impact criteria > (above) 95.3%: EP is more attractive overall than IMST. General conclusion: 0 < weight of impact criteria < 65.6 %: BSD is more attractive overall, Weight of impact = 65.6%: BSD and BCP are equally attractive, Weight of impact > 65.6%: BCP is more attractive overall, Weight of impact = 72.6%: BSD and DCAD are equally attractive, Weight of impact > 72.6%: DCAD is more attractive, Weight of impact = 95.3%: EP and IMST are equally attractive, Weight of impact > 95.3%: EP is more attractive. II. Sensitive Analysis for the Criteria 2 – Context Figures 2426 a sensitivity analysis results regarding the weight of the context criteria (Cnt). Page 35 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society v Figure 28. Sensitivity analysis on the weight of the criteria context Figure 29. Sensitivity analysis to the weight of the criteria context for the BCP and DCAD options Figure 30. Sensitivity analysis to the weight of the criteria context for the BCP and IMST options Page 36 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society vi The sensitivity analysis of the weight of the criteria 2 (context) from figures 28 to 30 revealed the following results: Weight of context criteria < (below) 9.3%: BCP is more attractive overall than DCAD, Weight of context criteria = 9.3%: BCP and DCAD are equally attractive, Weight of context criteria > (above) 9.3%: DCAD is more attractive overall than BCP. Weight of context criteria < (below) 100%: BCP is more attractive overall than IMST, Weight of context criteria = 100%: BCP and IMST are equally attractive. General conclusion: 0 < weight of context criteria < 100%: BSD is more attractive overall, Weight of context criteria = 100%: BCP and IMST are equally attractive. From the sensitivity analysis above, only a variation of 68.18% (from 31.82 to 100%) in the weight of context criteria will make another option (in this case, BCP and IMST) become as attractive as BSD. III. Sensitive Analysis for the Criteria 3 – Type of Event\ In the sensitivity analysis referring to the weight of criteria type of event (TE), we obtained the results expressed in figures 31, 32 and 33. Figure 31. Sensitivity analysis on the weight of the criteria type of event Page 37 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society vii Figure 32. Sensitivity analysis to the weight of the criteria type of event for the DCAD and IMST options Figure 33. Sensitivity analysis to the weight of the criteria type of event for the BCP and DCAD options The sensitivity analysis of the weight of the criteria 3 (type of event) from Figures 27 to 29 revealed the following results: Weight of type of event criteria < (below) 21.3%: DCAD is more attractive overall than BCP, Weight of type of event criteria = 21.3%: BCP and DCAD are equally attractive, Weight of type of event criteria > (above) 21.3%: BCP is more attractive overall than DCAD. Weight of type of event criteria < (below) 100%: DCAD is more attractive overall than IMST, Weight of type of event criteria = 100%: DCAD and IMST are equally attractive. General conclusion: Page 38 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society viii 0 < weight of type of event criteria < 100%: BSD is more attractive overall, Weight of type of event criteria = 100%: DCAD and IMST are equally attractive. From the sensitivity analysis above, a high variation of 97.73% (from 2.27 to 100%) in the weight of the type of event criteria will make another option (in this case, DCAD and IMST) become attractive as BSD. IV. Sensitive Analysis for the Criteria 4 – Spread of the Fake News In the sensitivity analysis referring to the weight of criteria spread of the fake news (SFN), we obtained the results expressed in figures 30, 31, 32 and 33. Figure 30: Sensitivity analysis on the weight of the criteria spread of the fake news Page 39 of 46 Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Information, Communication & Ethics in Society ix Figure 31: Sensitivity analysis of the weight of the criteria SFN for the BCP and DCAD options Figure 32: Sensitivity analysis of the weight of the criteria SFN for the IMST and DCAD options Figure 33: Sensitivity analysis to the weight of the criteria spread of the fake news for the IMST and BCP options Page 40 of 46Journal of Information, Communication & Ethics in Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60