Comparing Traditional and Graphical Risk Matrices: A Case Study in Healthcare

ABSTRACT

Risk and opportunity assessments are essential for decision‐making in complex systems such as healthcare and critical infrastructure. However, widely used tools like the risk matrix fail to explicitly capture uncertainty. This study presents the first empirical comparison between a traditional risk matrix and a previously proposed graphical method that visualizes uncertainty using two‐dimensional intervals. In a comprehensive survey, healthcare professionals assessed identical scenarios using both methods. The graphical approach yielded systematically different results, particularly in the estimation of probabilities, and revealed differences across occupational groups and infrastructure experience. These findings suggest that explicitly representing uncertainty may enhance the transparency and nuance of qualitative risk assessments, potentially addressing key limitations of conventional tools. Such approaches could support more reflective and differentiated decision‐making in high‐stakes environments.

1. Introduction

The assessment of risks and opportunities in complex systems, such as healthcare infrastructures, is a complex process that requires careful consideration of assumptions, estimates, and uncertainties to create a solid basis for informed decision‐making (Anes et al. 2020; Chen et al. 2025; Liberda and Sly 2023; Pasman et al. 2017; Renn et al. 2022). Uncertainties are inherent to qualitative assessments and can substantially influence both the outcomes and the resulting decisions (Abdo et al. 2017; Pasman and Rogers 2020; Whipple 1987). The risk matrix, that is, the graphical representation of risk by combining the estimated likelihood of occurrence and the potential severity of impact, is the most common method for the qualitative assessment of risks, particularly in domains where data are limited or expert judgment plays a central role. Figure 1 (left) shows an example of a typical 5 × 5 risk matrix. The red areas indicate high‐risk zones where the combination of high probability and severe impact suggests immediate action or mitigation.

FIGURE 1.

Traditional method versus graphical assessment method considering uncertainty.

The risk matrix provides a clear visual representation, is easy to communicate, and allows for prioritization of a list of risks (Krause et al. 2013; Lemmens et al. 2022). It can also be flexibly adapted to different scenarios by modifying the scale, size, or evaluation criteria, and it does not require complex calculations, which makes it easier to use in many areas. This widespread use is also reflected in international risk management standards such as ISO 31000:2018 (International Organization for Standardization 2018) and ISO 22301:2019 (International Organization for Standardization 2019), which emphasize the importance of transparent, structured, and uncertainty‐aware risk assessments in organizational and infrastructure‐related contexts. However, the classic risk matrix is limited by its inability to account for variabilities and uncertainties in the assessment of probabilities of occurrence and impacts/potential (Acebes, González‐Varona et al. 2024; L. A. Cox 2008, 2009; Duijm 2015). At the same time, several authors have raised fundamental concerns about the use of risk matrices as analytical tools, particularly when they are used to assign likelihood and impact values or to derive aggregated risk levels in qualitative assessments (Flage and Røed 2012; Goerlandt and Reniers 2016). Although these concerns are well founded, risk matrices continue to be widely used in practical settings. In many organizations, they function as de facto instruments for structured decision support, especially under regulatory and organizational frameworks. This ongoing tension between academic critique and practical reliance highlights the need for empirical investigation into how different matrix designs influence risk and opportunity assessments in real‐world application contexts. Despite their potential, alternative approaches (Abrahamsen and Aven 2011; Goerlandt and Reniers 2016; Rass et al. 2017; Riveiro et al. 2014) that explicitly account for uncertainty, such as visual methods using bubbles or rectangles to illustrate its magnitude, remain underutilized in practical risk assessment contexts. These methods offer the advantage of systematically integrating variability and uncertainty into the assessment process. The assessment of risks and opportunities associated with complex interdependencies, long‐term scenarios, and events in the distant future is often difficult and fraught with uncertainty, even for experts with specific experience in the respective field (Zio 2016). Current research on risk surveys lacks a comprehensive comparison of different survey methods (Aven 2013; D. C. Cox and Baybutt 1981; Glette‐Iversen et al. 2023; Xie et al. 2024). This gap motivated us to develop a study design that would enable such a comparison. Using an empirical approach, supported by a designated software tool, experts assessed the risks and opportunities of a complex, long‐term project with two different survey methods. This article aims to investigate how different evaluation methods (Figure 1, right diagram) influence risk and opportunity assessments, with a particular focus on the explicit representation of uncertainty in one of the methods. The study is based on a survey conducted in the field of critical infrastructure in the healthcare sector. In addition to examining the chosen methods themselves, the study also analyzes the assessments across various professional groups, as well as the evaluations of healthcare professionals with and without direct thematic ties to critical infrastructures (Pascarella et al. 2021). This focus is directly aligned with recent regulatory developments, particularly the Critical Entities Resilience Directive (EU 2022/2557) (European Union 2022), which requires critical infrastructure operators in EU member states to assess risks under consideration of uncertainty, interdependencies, and evolving threat landscapes. Section 3 explains the defined research questions and associated hypotheses in detail. The graphical method shown in Figure 2 builds on the uncertainty representation technique proposed by Duijm (2015), which extends the classic risk matrix by explicitly incorporating uncertainty into both dimensions. In this approach, the probability of occurrence is plotted on the ordinate and the impact or potential on the abscissa, as in traditional matrices.

FIGURE 2.

Explanation of the graphical assessment method considering uncertainty.

Rather than assigning a single point estimate, the experts draw a rectangle in the area, the height and width of which indicate the amount of certainty. The lower left corner of the rectangle marks the minimum estimated values for both dimensions, whereas the upper right corner represents the respective maxima. The vertical extent of the rectangle reflects uncertainty regarding the probability of occurrence, and the horizontal extent represents uncertainty regarding the impact. This technique allows for a more differentiated and transparent representation of uncertainty in qualitative risk assessments and supports decision‐makers in evaluating both the assessed risks and the associated confidence in that assessment. It is, however, important to note that this does not capture the underlying cause of the uncertainty or suggest any method to reduce it before or handle it during the decision‐making. We thus abstain from discussing the subsequent step of making decisions based on such data and confine ourselves to the comparison of the results of the discrete and the graphical methods (see Section 7). In this context, healthcare professionals with responsibilities related to critical infrastructure form a particularly relevant subgroup for risk assessment research. Their work environments are often marked by complex interdependencies, elevated stakes, and dynamic risk constellations, which may foster distinct patterns of risk perception and uncertainty awareness. Similarly, comparisons across occupational roles within the healthcare sector allow us to explore how different professional perspectives influence risk and opportunity assessments. These distinctions are not only theoretically meaningful but also practically relevant in light of the regulatory frameworks mentioned above (e.g., ISO 31000, ISO 14971, EN 15224), which emphasize the need for uncertainty‐aware risk assessments across all critical infrastructure sectors.

2. Related Work

Various established methods are available for qualitative risk and opportunity assessment, including risk matrices (L. A. Cox 2008), scenario analyses (Schoemaker 1995), Delphi methods (Okoli and Pawlowski 2004), bow‐tie analyses (de Ruijter and Guldenmund 2016; Omidvar et al. 2022), fuzzy logic (Whipple 1987), root cause analyses (Andersen and Fagerhaug 2006), and other approaches (Golfarelli et al. 2006; Kletz 2001; Pahl and Richter 2009). These methods are often used when quantitative data are not or are only insufficiently available or uncertainties are difficult to quantify. In such cases, there is a lack of reliable figures or measurable variables that would be required to precisely describe a risk or opportunity. The causes can be manifold, ranging from a lack of data and subjective assessments to high levels of complexity and variability caused by dynamic influencing factors, interactions, ambiguities, and human perceptions. One problem with the classic risk matrix is that it is based on fixed categories of impacts and likelihoods (Acebes, Curto et al. 2024; Duijm 2015), which are presented as crisp value(‐ranges) or qualitative ordinal classes (e.g., low, medium, high). These rigid structures limit the informative value, particularly in the case of uncertainties, as they cannot adequately depict bandwidths or dynamic developments (Krisper 2021; Wall 2011). This limitation becomes especially relevant in dynamic and complex environments, such as those that frequently occur in the area of critical infrastructure (Jahnel 2015; Thomas et al. 2013). To overcome these limitations, advanced methods can be used that take better account of uncertainties and their effects. Examples include interval‐based risk matrices, multidimensional matrices, or visualizations that enable uncertainty representations through color gradients, bandwidths, or additional axes (Dhungana et al. 2025; Goerlandt and Reniers 2016). These approaches allow a more transparent and flexible representation of uncertainties and their potential effects (Acebes, Curto et al. 2024; L. A. Cox 2009; Crotty and Daniel 2022; Kamal et al. 2021; Neff 2002; Zhang and Zhang 2022). Among these approaches, the graphical method introduced by Duijm (2015) and further specified by Rass et al. (2017) (Wachter et al. 2017) was selected for this study. This method was chosen because it directly visualizes uncertainty through a two‐dimensional rectangular area, maintaining the familiar structure of conventional risk matrices while extending their informational depth. Unlike many other uncertainty representation techniques, it does not require a complete shift in conceptual framework or visual logic, which makes it especially suitable for interdisciplinary use in healthcare and critical infrastructure contexts. In these settings, assessments are often based on expert judgment and must be communicated effectively across diverse professional backgrounds. Furthermore, critical infrastructure projects typically involve high levels of complexity, incomplete information, and evolving risk constellations. The graphical method accommodates these characteristics by allowing experts to express uncertainty explicitly and intuitively, without reducing it to a single point estimate. Its accessibility, compatibility with established tools, and ability to reflect real‐world uncertainty dynamics make it particularly appropriate for high‐stakes, interdisciplinary application domains such as healthcare and critical infrastructure. Despite the widespread use of the traditional risk matrix for qualitative risk and opportunity assessments, a significant research gap exists regarding its direct comparison with alternative methods under identical conditions. In particular, no prior scientific studies have systematically compared the classical risk matrix with an alternative risk assessment method by having the same participants assess identical scenarios using both methods at a time interval. Additionally, there is a lack of studies that statistically analyze the consistency and variability of these assessments, especially in complex contexts such as critical infrastructure.

This gap is particularly relevant, as the classical risk matrix is frequently criticized in the literature for its inability to adequately capture uncertainties and dynamic developments (Aven 2015; L. A. Cox 2008; Flage and Aven 2009; Leitch 2010; Saw et al. 2009). Advanced methods, such as interval‐based or multidimensional matrices, are often proposed as solutions, but their practical application and effectiveness in direct comparison to the classical matrix remain underexplored. The absence of such empirical research limits our understanding of the strengths and weaknesses of these methods and their potential to provide more reliable risk assessments. Moreover, both international standards and regulatory frameworks explicitly call for risk assessments that incorporate uncertainty and account for dynamic threat scenarios. Yet, practical tools that operationalize such recommendations remain limited. To address this gap, this study conducts a direct comparison between the classical risk matrix and an alternative graphical risk assessment method under controlled conditions. The experiment involves identical scenario assessments by the same participants for both methods, with results statistically analyzed to evaluate consistency, variability, and potential advantages. By focusing on applications in the healthcare sector, which is a critical infrastructure, this study provides novel insights into the practical implications of alternative methods for risk and opportunity assessment.

3. Research Question and Hypotheses

This study systematically compares two approaches to risk and opportunity assessments using distinct evaluation methods, illustrated through a survey in the healthcare sector. In addition to examining the methods themselves, the study analyzes the assessments of professionals across different occupational groups and independently evaluates differences based on experience with or without critical infrastructures. One of the methods under review enables the explicit representation of uncertainties related to both the impact/potential and the probability of occurrence. The results provide new insights into the method's applicability, which have not been explored in this way before. The study design involved all participants evaluating the same set of risk and opportunity scenarios using both assessment methods, separated by a fixed 2‐week interval. To structure the analysis, each research question was mapped to a corresponding hypothesis (see Table 1). As individual hypotheses cover multiple aspects—such as comparisons of impact, probability of occurrence, and subgroup differences—these were systematically structured into sub‐hypotheses. To ensure a more concise presentation, related aspects were grouped within the same hypothesis and tested using the same statistical method. Each hypothesis is split into specific sub‐hypotheses, which are listed in matrix format in Tables 2 and 3 but not described in detail in the text for reasons of brevity. Statistical analyses were conducted for all sub‐hypotheses of Hypotheses 1 through 10. These hypotheses assume equality of values, whereas their corresponding alternative hypotheses represent inequalities, meaning they were tested as two‐sided hypotheses. Table 2 provides an overview of all tested hypotheses comparing the traditional and graphical assessment methods across different evaluation dimensions. Table 3 summarizes the hypotheses related specifically to the graphical method with uncertainty representation. Each hypothesis is labeled as Hx.y, where x denotes the hypothesis group ID defined in Table 1, and y identifies the respective sub‐item within that group. The rows distinguish between outcome types (risk vs. opportunity) and assessment dimensions (impact/potential and probability of occurrence). The columns indicate the population subgroup to which the respective hypothesis applies: either the entire sample (all), specific occupational groups (e.g., administration, caregivers, management, medical professionals, technicians), or participants with or without a connection to critical infrastructure.

TABLE 1.

Generalized research questions and hypotheses applicable to both risk and opportunity assessments.

ID	Research question	Hypothesis
1	Do the traditional assessment method (risk or opportunity matrix) and the graphical assessment method (rectangle selection) yield the same results when evaluating risk or opportunity questions in terms of probability of occurrence and impact or potential?	It is assumed that the application of the traditional assessment method (risk or opportunity matrix) and the graphical assessment method (rectangle selection) will lead to identical results for the same questions when evaluating risks/opportunities concerning the impact/potential and the probability of occurrence
2	Do the traditional assessment method (risk or opportunity matrix) and the graphical assessment method (rectangle selection) yield the same results when evaluating risks and opportunities by different occupational groups, both in terms of the probability of occurrence and the impact or potential for the same questions?	It is assumed that the amount of coherence between the traditional and the graphical assessment method does not differ between occupational groups in the case of risk/opportunity assessment concerning the impact/potential and the probability of occurrence
3	Do the traditional assessment method (risk or opportunity matrix) and the graphical assessment method (rectangle selection) yield the same results when evaluating risks and opportunities by healthcare professionals with and without a connection to critical infrastructure, both in terms of the probability of occurrence and the impact or potential for the same questions?	It is assumed that the amount of coherence between the traditional and the graphical assessment method does not differ between healthcare professionals connected to critical infrastructure and those without such a connection in the case of risk/opportunity assessment concerning the impact/potential and the probability of occurrence
4	Do the traditional and graphical assessment methods produce identical results for impact, potential, and probability of occurrence in risk and opportunity assessments for healthcare professionals across different occupational groups, such as administration, caregivers, managers, medical professionals, and technicians?	It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the areas of risk assessment and opportunity assessment will lead to identical results regarding the impact, potential, and probability of occurrence of healthcare professionals across various occupational groups, including administration, caregivers, managers, medical professionals, and technicians
5	Do the traditional and graphical assessment methods produce identical results in terms of impact, potential, and probability of occurrence for healthcare professionals, regardless of their connection to critical infrastructure, when applied to identical questions in risk and opportunity assessments?	It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the areas of risk and opportunity assessment, concerning impact and potential, and the probability of occurrence, will lead to identical results for healthcare professionals with and without a connection to critical infrastructure
6	How does the estimation uncertainty of the impact compare to the estimation uncertainty of the probability of occurrence with the same scaling of the two variables?	For risks or opportunities assessed using the graphical survey method, the estimation uncertainty of the impact or potential is equal to the estimation uncertainty of the probability of occurrence, provided that both variables are measured on the same scale
7	How does the estimation uncertainty of impact or potential compare to the estimation uncertainty of probability across different occupational groups?	Regarding the graphical assessment method, the estimation uncertainty of impact is equal to the estimation uncertainty of probability for every professional group
8	Is the estimation uncertainty of impact or potential, when both quantities are on the same scale, equal to the estimation uncertainty of probability among healthcare professionals with or without a connection to critical infrastructure?	Regarding the graphical assessment method, the estimation uncertainty of impact, when both variables are measured on the same scale, is equal for healthcare professionals with and without a connection to critical infrastructure as the estimation uncertainty of probability
9	Does the estimation uncertainty of impact or potential and probability in the graphical assessment method remain consistent across different occupational groups?	Regarding the graphical assessment method, the estimation uncertainty of impact and probability does not differ between occupational groups
10	Does the estimation uncertainty of impact, potential, and probability in the graphical assessment method remain equally significant among healthcare professionals with and without a connection to critical infrastructure?	Regarding the graphical assessment method, the estimation uncertainty of impact and probability does not differ between healthcare professionals with and without a connection to critical infrastructure

TABLE 2.

Hypotheses overview for “traditional versus graphical method.”

TABLE 3.

Hypotheses overview for the “graphical method with uncertainty.”

For illustration, Hypothesis H1.2 examines whether assessments of risk probability differ between the traditional and graphical methods across the full sample (see Table 2). Hypothesis H9.2, in turn, investigates whether participants with and without a background in critical infrastructure differ in their uncertainty estimates of risk probability using the graphical method (see Table 3).

In accordance with ISO 31000 and the glossary of the Society for Risk Analysis (SRA), the term risk is commonly defined as the effect of uncertainty on objectives, encompassing both positive and negative outcomes. However, for the purpose of this study, risks and opportunities are treated as analytically distinct constructs. This distinction follows the structure of the applied assessment instrument, which separates negative (risk‐related) and positive (opportunity‐related) scenarios both semantically and in terms of their evaluative dimensions. Specifically, the underlying instrument provides tailored descriptors for impact and probability in risk contexts and for potential and probability in opportunity contexts. Accordingly, the terminology used in the hypotheses reflects the structure of the assessment instrument: Impact denotes the estimated magnitude of a negative outcome in risk scenarios, potential refers to the estimated magnitude of a positive outcome in opportunity scenarios, and probability captures the estimated likelihood of occurrence across both contexts. These terms are operationalized through the graphical method shown in Figure 2, where respondents express both impact or potential and probability of occurrence by selecting two‐dimensional rectangles. This format enables the visualization of uncertainty as ranges rather than fixed point estimates. On the basis of this conceptual framing, the following research questions and hypotheses were developed.

4. Methods

4.1. Materials

This study was conducted as part of a long‐term, large‐scale hospital reorganization project involving several medical departments. The project comprises extensive construction activities and comprehensive upgrades to the hospital's technical infrastructure, with the overarching goal of modernizing the physical environment and optimizing organizational processes. The risk and opportunity scenarios used in this study were specifically developed for this project by the hospital's risk manager, based on internal project documentation and expert input. A total of 68 distinct risk scenarios and 18 opportunity scenarios were defined and categorized into the following overarching domains: environment and ecology, legal aspects, participants and affected parties, project consequences, project environment, project organization, project process, project specifications, safety, technology, and unforeseeable events. All scenarios were presented in a standardized written format.

4.2. Participant Roles and Survey Scope

Survey participants represented a range of relevant professional domains, including administration, caregiving, clinical medicine, healthcare management, and technical services. All had substantial experience in complex hospital redevelopment projects and were familiar with the specific rating scales used to assess both the potential impact and the probability of occurrence of each scenario. This aimed to promote a consistent understanding of both the severity of consequences and the frequency‐based probability of occurrence and provided a shared evaluative framework across all participants. Participants were asked to evaluate two distinct dimensions for each scenario, using different rating scales depending on the assessment method applied and whether the scenario represented a risk or an opportunity. For example, one risk scenario in the category project specifications described the “risk of key positions lacking identification with the project and its objectives, potentially leading to insufficient commitment, resistance, or inefficient decision‐making.” Another example described a scenario from the domain unforeseeable events: “Unpredictable delays, disruptions or complete failures of supply chains and essential services can be triggered by external factors such as geopolitical crises, natural disasters, economic instability, cyberattacks or pandemics. These events can have a significant impact on the project by compromising the schedule, driving up costs and limiting the availability of key resources.”

The study was embedded in an internal project of the Landeskrankenanstalten‐Betriebsgesellschaft—KABEG (www.kabeg.at), and explicit permission was granted to publish the results for scientific purposes. A total of 4667 paired assessments for risk scenarios and 1199 paired assessments for opportunity scenarios were obtained from the project‐related survey. To enable subgroup analyses, participants were categorized according to their professional roles and their experience with critical infrastructure contexts. The sample included professionals from administrative, clinical, caregiving, technical, and managerial roles, each contributing a distinct perspective on risk and opportunity assessments. In addition, participants were classified on the basis of whether their current professional responsibilities involved critical infrastructure systems. This allowed for comparative analyses between healthcare professionals with and without such experience, addressing the question of whether exposure to complex, high‐stakes environments influences the perception and handling of uncertainty.

4.3. Experimental Design

The study was designed to compare two methods for assessing risk and opportunities within the context of a real‐world healthcare infrastructure project. Participants were randomly assigned to two balanced groups. Group I began the evaluation with the traditional assessment method, whereas group II started with the graphical, uncertainty‐based format. Exactly 14 days after completing the first round, each participant received the same set of scenarios again, this time evaluated using the respective alternative method. The interval was deliberately chosen to minimize recall bias and support independent responses. Participants were asked to evaluate the impact and the probability of occurrence of each scenario. To avoid subjective ambiguities in the understanding of terms along which impacts and likelihoods were rated (“low,” “medium,” etc.), we provided the same definitions of terms in all experiments and both types of assessments.

They were allowed to skip individual items if they considered them outside their area of expertise. These skipped items were automatically excluded from the second round of data collection, ensuring that only relevant and assessable scenarios were presented again. Upon completion of both assessment rounds, participants were also asked to reflect on the two methods and provide structured feedback regarding usability, clarity, and perceived advantages or limitations. Figure 3 provides a visual overview of the survey procedure.

FIGURE 3.

Survey procedure.

To ensure clarity and content validity, all scenarios were iteratively reviewed and refined in collaboration with internal experts from relevant domains, including legal, technical, and clinical departments. The survey was implemented using a secure purpose‐built online platform that complied with GDPR and ethical standards. This platform supported both assessment methods, facilitated informed consent procedures, and enabled structured participant management. Given the conceptual and structural differences between risks and opportunities, their results were analyzed separately. A complete listing and classification of all scenarios is available in the supplementary material.

A paired assessment refers to the evaluation of the same scenario by the same individual using both the traditional and the graphical method.

To test the hypotheses presented in Tables 1–3, paired assessments from each participant were assigned to one of five occupational groups. For additional analyses, assessments were also classified on the basis of whether the participant had a direct connection to critical infrastructure. The term number of overlaps used in the subsequent analysis refers to cases in which the result of the traditional method matches the converted result of the graphical method, as described in Section 4.1.

The study received prior approval from the Ethics Committee of the State of Carinthia, Austria. Before participating, all individuals were informed about the anonymous use of their data and provided explicit consent by confirming a declaration on the introductory page of the online survey. The survey was conducted using a secure, custom‐developed software platform that supported both assessment methods and participant management.

No minors or patients were included in the survey. The recruitment period lasted from August 17, 2022, to December 9, 2022, ensuring sufficient participation to meet the study's objectives. After the completion of data collection, the research team concentrated on refining aggregation approaches and performing comprehensive analyses to support the robustness and validity of the findings.

4.4. Statistical Analysis

To enable a comparison between the traditional and graphical assessment methods, the graphical assessments were transformed into discrete values by calculating the center point of the respective rectangles. These values were then mapped onto the same 5 × 5 scale as used in the traditional matrix as shown in Figure 4, enabling coherence analysis between methods.

FIGURE 4.

Conversion of the graphical assessment result for comparison with the traditional values.

All statistical analyses were conducted using R (Version 4.4.2). The significance level for all tests was set at α = 0.05.

4.4.1. Binomial Test

To evaluate whether the two assessment methods yield identical results for the same questions (H1.1–H1.4), we used binomial tests (Siegel and Castellan Jr. 1988) to assess whether the observed agreement differs significantly from what would be expected by chance.

4.4.2. Chi‐Square Test

To examine whether the coherence between the two assessment methods varies across occupational groups (H2.1–H2.4) and between healthcare professionals with and without a connection to critical infrastructure (H3.1–H3.4), we applied chi‐square tests (χ ² test) of independence (Agresti 2013), which test the association between categorical variables.

4.4.3. Paired Wilcoxon Signed‐Rank Test

The Paired Wilcoxon Signed‐Rank Test (Wilcoxon 1945) was used to compare matched observations between the traditional and graphical assessment methods (H4.1–H4.20, H5.1–H5.8). As a non‐parametric alternative to the paired t‐test, it is appropriate for ordinal or non‐normally distributed data. The test was also used to compare estimation uncertainty between probability of occurrence and impact/potential within individuals (H6.1–H8.4), both across and within professional groups.

4.4.4. Kruskal–Wallis Test

To examine whether estimation uncertainty differs between occupational groups (H9.1–H9.4), we used the Kruskal–Wallis test (Kruskal and Wallis 1952). This non‐parametric alternative to one‐way ANOVA is suitable for comparing multiple independent groups for ordinal responses or when normality cannot be assumed. It was applied separately to risk and opportunity assessments for both impact/potential and probability of occurrence.

4.4.5. Dunn Test With Bonferroni Correction

Following the Kruskal–Wallis test, we conducted a Dunn test with Bonferroni correction (Dinno 2015; Dunn 1964) (Shaffer 1995) to identify specific group differences in pairwise comparisons. This post hoc procedure aligns with the rank‐based logic of the Kruskal–Wallis test and controls for Type I error inflation due to multiple testing.

4.4.6. Mann–Whitney U‐Test

To test whether estimation uncertainty differs between healthcare professionals with and without a connection to critical infrastructure (H10.1–H10.4), we applied the Mann–Whitney U‐Test (Mann and Whitney 1947). As a non‐parametric alternative to the independent t‐test, it is appropriate for comparing two independent groups with potentially non‐normal data. The analysis was conducted separately for risk and opportunity assessments.

4.5. Demographic and Professional Characteristics of the Sample

Of the N = 104 individuals invited to participate, 96 initiated the assessment at Time Point 1 (T1), and n = 75 completed the follow‐up assessment at Time Point 2 (T2), yielding a sample attrition rate of approximately 22% between T1 and T2. The sample included 31 female and 44 male participants. Information on age was available for all 75 participants, with a mean age of M = 47.89 years (SD = 10.58, range: 23–66). In terms of professional background, participants were employed in various roles within the healthcare sector, including medical (n = 8), technical (n = 30), administrative (n = 19), caregiving (n = 11), and management positions (n = 7). Educational levels varied among participants, with the majority holding a vocational degree (n = 27), followed by bachelor's (n = 4) and master's degrees (n = 44). Data on employment duration were available for all individuals, with an average of 22.46 (SD = 9.91).

5. Results

The data used for further analysis comes from a survey conducted as part of an extensive project in the healthcare sector.

5.1. Hypothesis 1: Comparison of Assessment Method Results for Identical Questions

Table 4 summarizes the hypotheses comparing the traditional and graphical methods for identical scenarios. The corresponding statistical results are shown in Table 5.

TABLE 4.

Summary of hypotheses in hypothesis group 1.

It is assumed that the application of both the traditional assessment method and the graphical assessment method will lead to identical results for the same questions when evaluating	risks concerning the impact	H1.1
	risks concerning the probability of occurrence	H1.2
	opportunities concerning the potential	H1.3
	opportunities concerning the probability of occurrence	H1.4

TABLE 5.

Results of the statistical tests of hypothesis group 1.

Note: All four hypotheses (H1.1–H1.4) were rejected on the basis of the statistical tests. The results show that the traditional and the graphical assessment methods yield significantly different outcomes in terms of risk and opportunity evaluation. The differences were observed in both impact and probability of occurrence for risks, as well as in potential and probability of occurrence for opportunities.

5.2. Hypothesis 2: Comparison of Assessment Method Results Across Occupational Groups

Table 6 presents the hypotheses of group 2, which test whether the coherence between the two assessment methods differs across occupational groups. The results of the corresponding statistical tests are summarized in Table 7.

TABLE 6.

Summary of hypotheses in hypothesis group 2.

It is assumed that the amount of coherence between the traditional and the graphical assessment method does not differ between occupational groups in the case of	risk assessment concerning the impact	H2.1
	risk assessment concerning the probability of occurrence	H2.2
	opportunity assessment concerning the potential	H2.3
	opportunity assessment concerning the probability of occurrence	H2.4

TABLE 7.

Results of the statistical tests of hypothesis group 2.

Note: In summary, the findings indicate that occupational groups differ significantly in their coherence between the graphical and the traditional method. This is the case in the assessments of risk (both impact and probability of occurrence) and in the assessment of opportunity probability as well. There is more similarity in the coherence across groups in the case of opportunity potential. The medical professionals showed the least coherence in terms of both impact and probability of occurrence. In contrast, the groups of managers and technicians showed a higher degree of coherence, particularly in the assessment of risk impact. All hypotheses, except for H2.3, are rejected, indicating that the amount of coherent does not differ between both professional groups, with the exception of risk probability.

5.3. Hypothesis 3: Comparison of Assessment Method Results Across Professionals in Critical Infrastructure Contexts

Table 8 presents the hypotheses of group 3, which test whether the coherence between the two assessment methods differs depending on whether participants are connected to critical infrastructure. The corresponding statistical results are shown in Table 9.

TABLE 8.

Summary of hypotheses in hypothesis group 3.

It is assumed that the amount of coherence between the traditional and the graphical assessment method does not differ between healthcare professionals connected to critical infrastructure and those without such a connection in the case of	risk assessment concerning the impact	H3.1
	risk assessment concerning the probability of occurrence	H3.2
	opportunity assessment concerning the potential	H3.3
	opportunity assessment concerning the probability of occurrence	H3.4

TABLE 9.

Results of the statistical tests of hypothesis group 3.

Note: The results indicate no significant differences in the coherence of assessments between healthcare professionals with and without a connection to critical infrastructure for risk impact, opportunity potential, and opportunity probability of occurrence (H3.1, H3.3, and H3.4). However, there is a significant difference in coherence between the graphical and traditional methods regarding the risk probability of occurrence (H3.2). This finding suggests that experience or involvement in critical infrastructure may influence risk perception. In conclusion, all hypotheses except H3.2 are retained, indicating that both professional groups provide largely consistent assessments, with the exception of risk probability.

5.4. Hypothesis 4: Comparison of Assessment Methods Across Occupational Groups

Table 10 lists the hypotheses of group 4, which examine whether the two assessment methods yield consistent results within each occupational group. The statistical results are presented in Table 11.

TABLE 10.

Summary of hypotheses in hypothesis group 4.

It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the area of risk assessment will lead to identical results	regarding the impact for the occupational group	Administration	H4.1
		Caregiver	H4.5
		Management	H4.9
		Medical	H4.13
		Technician	H4.17
	regarding the probability of occurrence for the occupational group	Administration	H4.2
		Caregiver	H4.6
		Management	H4.10
		Medical	H4.14
		Technician	H4.18
It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the area of opportunity assessment will lead to identical results	regarding the potential for the occupational group	Administration	H4.3
		Caregiver	H4.7
		Management	H4.11
		Medical	H4.15
		Technician	H4.19
	regarding the probability of occurrence for the occupational group	Administration	H4.4
		Caregiver	H4.8
		Management	H4.12
		Medical	H4.16
		Technician	H4.20

TABLE 11.

Results of the statistical tests of hypothesis group 4.

Note: The results of the statistical tests for hypothesis group 4 reveal significant differences between the traditional and graphical assessment methods across different occupational groups. For risk impact, most occupational groups show no significant differences in the results from the two methods, except for medical and technician groups, where significant differences were found. For risk probability of occurrence, all occupational groups show significant differences between the two methods, indicating that the traditional and graphical methods yield different results when assessing the probability of occurrence of risks. For opportunity potential, there are no significant differences between the two methods for most occupational groups, except for caregivers. For opportunity probability of occurrence, significant differences were found across all occupational groups. In summary, although there is some consistency in the results for opportunity potential and risk impact across occupational groups, the probability of occurrence assessments, both for risks and opportunities, show significant differences between the traditional and graphical methods for all groups.

5.5. Hypothesis 5: Comparison of Assessment Methods Across Healthcare Professionals in Critical Infrastructure Contexts

Table 12 outlines the hypotheses of group 5, which test whether the two assessment methods yield consistent results within each subgroup of professionals, depending on their connection to critical infrastructure. The corresponding statistical results are shown in Table 13.

TABLE 12.

Summary of hypotheses in hypothesis group 5.

It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the area of risk assessment will lead to identical results	regarding the impact for healthcare professionals	with connection to critical infrastructure	H5.1
		without connection to critical infrastructure.	H5.5
	regarding the probability of occurrence for healthcare professionals	with connection to critical infrastructure	H5.2
		without connection to critical infrastructure	H5.6
It is assumed that applying the traditional assessment method and the graphical assessment method to identical questions in the area of opportunity assessment will lead to identical results	regarding the potential of opportunities for healthcare professionals	with connection to critical infrastructure	H5.3
		without connection to critical infrastructure	H5.7
	regarding the probability of occurrence for healthcare professionals	with connection to critical infrastructure	H5.4
		without connection to critical infrastructure	H5.8

TABLE 13.

Results of the statistical tests of hypothesis group 5.

Note: For healthcare professionals connected to critical infrastructure, both assessment methods produced significantly different results in most cases, except for the opportunity potential (H10.3), where the difference was not statistically significant. For healthcare professionals without connection to critical infrastructure, both methods also showed significant differences, except for risk impact (H5.5), where the methods produced similar results. The findings suggest that the traditional and graphical assessment methods do not lead to identical results for most risk and opportunity assessments concerning healthcare professionals, especially in contexts related to critical infrastructure.

5.6. Hypothesis 6: Estimation Uncertainty in Risk and Opportunity Assessments

Table 14 summarizes the hypotheses of group 6, which test whether estimation uncertainty differs between impact (or potential) and probability of occurrence when using the graphical method. The corresponding statistical results are shown in Table 15.

TABLE 14.

Summary of hypotheses in hypothesis group 6.

For risks assessed using the graphical survey method, the estimation uncertainty of the impact is equal to the estimation uncertainty of the probability of occurrence, provided that both variables are measured on the same scale	H6.1
For opportunities assessed using the graphical survey method, the estimation uncertainty of the potential is equal to the estimation uncertainty of the probability of occurrence, provided that both variables are measured on the same scale	H6.2

TABLE 15.

Results of the statistical tests of hypothesis group 6.

Note: The results reveal significant differences in estimation uncertainty between impact/potential and probability of occurrence for both risks and opportunities, with impact or potential showing consistently higher uncertainty than probability of occurrence. Consequently, both hypotheses (H6.1 and H6.2) are rejected, confirming that estimation uncertainty varies depending on whether impact/potential or probability of occurrence is assessed, even when both variables are measured on the same scale.

5.7. Hypothesis 7: Estimation Uncertainty Concerning Occupational Groups

Table 16 lists the hypotheses of group 7, which examine whether estimation uncertainty differs between impact (or potential) and probability of occurrence within each occupational group. The statistical results are presented in Table 17.

TABLE 16.

Summary of hypotheses in hypothesis group 7.

In the risks of the graphical survey method, the estimation uncertainty of impact, on the same scale for both quantities across each occupational group, is equal to the estimation uncertainty of probability of occurrence for the occupational group	Administration	H7.1
	Caregiver	H7.3
	Management	H7.5
	Medical	H7.7
	Technician	H7.9
In the opportunities of the graphical survey method, the estimation uncertainty of potential, on the same scale for both quantities across each occupational group, is equal to the estimation uncertainty of probability of occurrence for the occupational group	Administration	H7.2
	Caregiver	H7.4
	Management	H7.6
	Medical	H7.8
	Technician	H7.10

TABLE 17.

Results of the statistical tests of hypothesis group 7.

Note: The results indicate significant differences in estimation uncertainty between impact/potential and probability of occurrence across all occupational groups for both risk and opportunity assessments. In every occupational group (administration, caregiver, management, medical, and technician), the uncertainty in estimating impact/potential was consistently higher than the uncertainty in estimating probability of occurrence. As a result, all hypotheses (H7.1–H7.10) are rejected, demonstrating that estimation uncertainty varies depending on the factor being assessed (impact/potential vs. probability of occurrence) across all occupational groups, even when both are measured on the same scale.

5.8. Hypothesis 8: Estimation Uncertainty Concerning Healthcare Professionals in Critical Infrastructure Contexts

Table 18 presents the hypotheses of group 8, which examine whether estimation uncertainty differs between impact (or potential) and probability of occurrence depending on whether professionals are connected to critical infrastructure. The corresponding results are shown in Table 19.

TABLE 18.

Summary of hypotheses in hypothesis group 8.

In the context of risks assessed using the graphical survey method, the estimation uncertainty of impact, on the same scale for both quantities, is equal to the estimation uncertainty of probability for healthcare professionals	with a connection to critical infrastructure	H8.1
	without a connection to critical infrastructure	H8.3
In the context of opportunities assessed using the graphical survey method, the estimation uncertainty of potential, on the same scale for both quantities, is equal to the estimation uncertainty of probability for healthcare professionals	with a connection to critical infrastructure	H8.2
	without a connection to critical infrastructure	H8.4

TABLE 19.

Results of the statistical tests of hypothesis group 8.

Note: The results reveal significant differences in estimation uncertainty between impact/potential and probability of occurrence across healthcare professionals with and without a connection to critical infrastructure. In both risk and opportunity assessments, healthcare professionals with a connection to critical infrastructure showed lower estimation uncertainty in impact/potential compared to probability of occurrence. Similarly, healthcare professionals without a connection to critical infrastructure also demonstrated higher uncertainty for impact/potential than for probability of occurrence. All hypotheses (H8.1–H8.4) are rejected, as the findings indicate that the estimation uncertainty for impact/potential is consistently higher than for probability of occurrence, regardless of whether the professionals are connected to critical infrastructure.

5.9. Hypothesis 9: Estimation Uncertainty Across Occupational Groups

Table 20 outlines the hypotheses of group 9, which test whether estimation uncertainty differs across occupational groups for both impact/potential and probability of occurrence. The corresponding results are shown in Table 21.

TABLE 20.

Summary of hypotheses in hypothesis group 9.

In the context of risks assessed using the graphical survey method, the estimation uncertainty does not differ between occupational groups regarding	the impact	H9.1
	the probability of occurrence	H9.2
In the context of opportunities assessed using the graphical survey method, the estimation uncertainty does not differ between occupational groups regarding	the potential	H9.3
	the probability of occurrence	H9.4

TABLE 21.

Results of the statistical tests of hypothesis group 9.

Note: The results indicate significant differences in estimation uncertainty for both impact/potential and probability of occurrence across all occupational groups. For risk assessments, the estimation uncertainty for both impact and probability of occurrence differs significantly between the occupational groups, with the highest uncertainty observed in the medical profession. Similarly, for opportunity assessments, significant differences were found in the estimation uncertainty for both potential and probability of occurrence across the occupational groups. Thus, all hypotheses (H9.1–H9.4) are rejected, demonstrating that estimation uncertainty differs significantly between occupational groups, suggesting that different professions perceive and assess risks and opportunities with varying levels of certainty. Given these significant differences, a post hoc test is required to determine which specific groups differ from one another.

By applying the Dunn test, we can determine that occupational groups significantly differ in their estimation uncertainties, providing more granular insights into how different professional groups assess risks and opportunities and where key differences in uncertainty arise. This will provide clearer insights into how different professional groups approach risk and opportunity assessments and where key differences in uncertainty lie. For example, although the Kruskal–Wallis test has shown that medical professionals exhibit the highest uncertainty in impact estimation, the Dunn test will help confirm whether this uncertainty is significantly higher than that of other groups, such as administration or caregivers. The results from the Dunn test (Table 22) will highlight specific pairs of occupational groups with significant differences, and these findings will be discussed to improve decision‐making processes and address disparities in risk perception across professional roles.

TABLE 22.

Results of the Dunn test with Bonferroni correction.

Note: The results show significant differences in several pairwise comparisons between the occupational groups. In both the risk and opportunity assessment, the administrative and medical occupational groups showed the greatest differences in the assessment of impact, potential, and probability of occurrence. This indicates that the medical professional group perceives risks significantly differently than the administrative group. Other notable findings include the differences between administration and nursing staff, with the results of the test indicating significant differences in risk assessment, and between administration and technicians, where similar trends are observed. Overall, the results show clear differences between the occupational groups in terms of their risk perception and their assessment of opportunities. This could be due to different responsibilities, working conditions, and career prospects. The Bonferroni correction confirmed the statistical significance of most pairwise comparisons, which supports the robustness of the results.

5.10. Hypothesis 10: Estimation Uncertainty Across Professionals in Critical Infrastructure Contexts

Table 23 summarizes the hypotheses of group 10, which test whether estimation uncertainty differs between healthcare professionals with and without a connection to critical infrastructure. The corresponding statistical results are presented in Table 24.

TABLE 23.

Summary of hypotheses in hypothesis group 10.

In the context of risks assessed using graphical survey methods, the estimation uncertainty does not differ between healthcare professionals with and without a connection to critical infrastructure regarding	the impact	H10.1
	the probability of occurrence	H10.2
In the context of opportunities assessed using graphical survey methods, the estimation uncertainty does not differ between healthcare professionals with and without a connection to critical infrastructure regarding	the potential	H10.3
	the probability of occurrence	H10.4

TABLE 24.

Results of the statistical tests of hypothesis group 10.

Note: The results of the statistical tests show significant differences in the estimation uncertainty between healthcare professionals with and without a connection to critical infrastructure. The statistical tests revealed significant differences in estimation uncertainty for both risk and opportunity assessments between healthcare professionals with and without connections to critical infrastructure. In all cases, professionals connected to critical infrastructure exhibited lower uncertainty for impact and potential but higher uncertainty for probability of occurrence.

6. Discussion

The discussion follows the hypothesis groups to ensure a systematic reflection on each key dimension and its implications for applied risk assessment practices in healthcare. In this study, risks and opportunities are treated as analytically distinct categories. Although ISO 31000 and the SRA define risk in a broad sense as the effect of uncertainty on objectives, encompassing both positive and negative outcomes, many applied standards require an explicit consideration of both risks and opportunities. In healthcare, ISO 9001:2015 is widely used and demands a structured approach to identifying and addressing risks and opportunities separately to ensure system effectiveness and continual improvement. In practice, this often leads to a procedural separation. The assessment tools used in this study also differentiate systematically between risk and opportunity scenarios. They employ distinct scaling logics and terminology, especially in the evaluation of impact for risks and potential for opportunities. Accordingly, the methodological separation applied in this study reflects both normative practice in healthcare governance and the structure of the instruments used. It serves to ensure conceptual clarity in the interpretation of the findings. This conceptual and methodological framing provides the basis for the following discussion of results.

6.1. Hypothesis Group 1: Results for Identical Questions

The rejection of all hypotheses in group 1 indicates that the choice of assessment method substantially influences evaluation outcomes. We interpret this divergence as stemming from differences in cognitive engagement. The graphical method, by explicitly visualizing uncertainty, appears to encourage more deliberate reflection on potential variations in impacts and probabilities of occurrence. This interpretation is supported by findings from risk perception research and behavioral decision theory (Riveiro et al. 2014) (Bodenberger and Thommes 2025) (Kahneman and Tversky 1979), which shows that framing and visualization of uncertainty can shape how risks are evaluated. In contrast, the traditional risk matrix often promotes more intuitive judgments, potentially leading to oversimplification and underestimation of uncertainty (Padilla et al. 2021; Romeike and Hager 2020). These results suggest that the observed methodological differences reflect distinct cognitive processes triggered by each method. The graphical method's explicit emphasis on uncertainty may foster deeper cognitive engagement and support more differentiated evaluations of risks and opportunities, potentially explaining the systematic differences observed between the two approaches. Although these findings point to method‐related effects, they should not be interpreted solely as a consequence of matrix design. It is equally plausible that perceptual and cognitive factors, such as individual interpretation of scales or varying decision heuristics, contribute to the observed divergence. The implications of these factors are further addressed in Section 7.

6.2. Hypothesis Group 2: Results Across Occupational Groups

The overwhelming rejection of the hypotheses in hypothesis group 2 underscores the significant influence of occupational background on the coherence between traditional and graphical assessment methods. Although assessments of opportunity potential showed relatively consistent coherence, notable discrepancies emerged in risk assessments, particularly for impact and probability of occurrence. “Management” and “technicians” display higher coherence, likely due to their familiarity with structured analysis and data‐driven decision‐making. In contrast, “medical” professionals, who typically rely more on intuitive judgment and qualitative reasoning, showed lower coherence. These differences may reflect distinct cognitive styles and occupational cultures, as supported by previous research on profession‐specific risk perception (Slovic 2013).

6.3. Hypothesis Group 3: Results Across Professionals With and Without Connection to Critical Infrastructure

The results of hypothesis group 3 indicate that, in most cases, a professional connection to critical infrastructure does not significantly influence the coherence between traditional and graphical assessments. A notable exception emerged in the assessment of risk probability, where professionals with infrastructure responsibilities demonstrated significantly higher coherence.

This may reflect their heightened sensitivity to systemic risks and cascading effects, supported by risk perception research (Slovic 2013), which shows that direct exposure and specific expertise enable more differentiated and accurate probabilistic judgments. Nonetheless, overall coherence remained low in both groups, suggesting that other factors such as methodological limitations, cognitive styles, or differing interpretations of uncertainty may also contribute.

6.4. Hypothesis Group 4: Methods Across Occupational Groups

The results of hypothesis group 4 reveal significant differences in how occupational groups apply the traditional and graphical assessment methods, particularly in the evaluation of probabilities. In contrast, assessments of opportunity potential and risk impact were more consistent. These discrepancies may reflect differences in cognitive styles and training, especially regarding probabilistic reasoning and the interpretation of visual uncertainty. Notably, variations in the assessment of risk impact by “medical” and “technician” groups, and in opportunity potential by “caregivers,” may indicate professional biases toward either qualitative or quantitative approaches.

6.5. Hypothesis Group 5: Methods Across Healthcare Professionals With and Without Connection to Critical Infrastructure

The results of hypothesis group 5 show that traditional and graphical assessment methods often lead to significantly different results in both groups of healthcare professionals, independent of critical infrastructure experience. Exceptions include opportunity potential (for professionals with infrastructure roles) and risk impact (for those without), where the methods produced similar outcomes. This suggests that differences are more likely due to characteristics of the methods themselves than to professional experience alone. Specific patterns, however, point to a more nuanced relationship between context and experience.

Professionals with infrastructure responsibilities may be more sensitive to uncertainty, particularly regarding risk probability, possibly due to regular exposure to systemic risks and cascading effects. In contrast, those without responsibilities may assess risk impact more consistently, as impact is often more tangible and less dependent on probabilistic reasoning. These findings align with the previous research (Joslyn and LeClerc 2013; Kahneman and Tversky 1979; Slovic 2013) showing that exposure to complex systems influences how uncertainty is evaluated. Moreover, the visual communication of uncertainty has been shown to significantly affect risk estimates and decision‐making (Bodenberger and Thommes 2025; Hullman 2019).

6.6. Hypothesis Group 6: Estimation Uncertainty in Risk and Opportunity Assessments

The results of hypothesis group 6 show significant differences in estimation uncertainty between impact and probability of occurrence in risk assessments and between potential and probability of occurrence in opportunity assessments. This suggests that respondents engage differently with these two dimensions, even when measured on the same scale. The consistently higher uncertainty for impact and potential, compared to probability of occurrence, indicates that estimating the magnitude of outcomes is more difficult than judging their likelihood. This may be due to the abstract, context‐dependent nature of impact‐related judgments, which involve complex consequences and systemic effects.

In contrast, probability estimates often rely on heuristics or familiar reference points. These findings align with previous research showing that users tend to emphasize severity over likelihood, often prioritizing rare but extreme events over more probable scenarios with milder consequences (Pascarella et al. 2021). This bias, described as a form of risk aversion, can lead to misaligned prioritization in risk management, particularly when uncertainty is not explicitly addressed or communicated (Joint Task Force Transformation Initiative 2012; Kheybari et al. 2024).

The graphical method used in this study allows for range‐based estimates and highlights that uncertainty is not evenly distributed across the dimensions. Ignoring such variation, as in rigid matrix‐based approaches, may obscure important nuances. Therefore, tools that enable explicit representation of uncertainty may help users differentiate between types of uncertainty related to impact and probability.

In addition to highlighting differences in uncertainty, the graphical method also offers practical advantages for risk governance. By visualizing the range and spread of expert estimates, it allows decision‐makers to identify areas of high ambiguity or disagreement that might otherwise remain hidden in conventional risk matrices. This added layer of transparency can help to prioritize discussion, search for the underlying sources of uncertainty (to reduce or otherwise handle it), allocate resources more effectively, and foster shared understanding across professions. Although the method does not determine better decisions per se, it contributes information that may be highly relevant for reflective, well‐informed decision‐making in complex healthcare contexts.

6.7. Hypothesis Group 7: Estimation Uncertainty Concerning Occupational Groups

The results of hypothesis group 7 show that estimation uncertainty was consistently higher for impacts and potentials than for probabilities across all occupational groups, regardless of whether risks or opportunities were assessed. This suggests that the nature of the factor being evaluated has a substantial influence on the level of uncertainty. The greater uncertainty in assessing impact and potential likely stems from their abstract, interpretive nature, which contrasts with the more concrete estimation of probabilities. Notably, medical professionals exhibited the highest overall uncertainty, possibly due to the complexity and unpredictability of clinical decision‐making in high‐stakes environments. These findings are supported by research on risk perception (Kahneman et al. 2022; Qiu et al. 2024; Sundh 2024), which shows a tendency to overemphasize severe consequences, even when their probability is low. This may contribute to the elevated uncertainty in medical assessments.

Prior work (Pascarella et al. 2021) has criticized risk matrices in healthcare for overemphasizing consequences, leading to misaligned prioritizations. Methodological factors may further influence these results. The graphical method allows for explicit expression of uncertainty but also requires users to correctly interpret visual representations. Professionals less familiar with such tools may indicate broader ranges or struggle with scaling. In summary, estimation uncertainty appears to be shaped both by the type of judgment and by profession‐specific cognitive tendencies. This underscores the importance of discipline‐sensitive communication and targeted training to improve the consistency and reliability of risk assessments in multi‐professional settings.

6.8. Hypothesis Group 8: Estimation Uncertainty Concerning Healthcare Professionals With and Without Context to Critical Infrastructure

The results of hypothesis group 8 confirm the pattern observed throughout this study that estimation uncertainty is significantly higher for impact and potential than for probability of occurrence, regardless of whether respondents are affiliated with critical infrastructure. This suggests that assessing consequences poses greater cognitive demands than estimating probabilities, which often draws on more concrete, experience‐based heuristics. Healthcare professionals with a connection to critical infrastructure showed slightly lower levels of uncertainty, possibly reflecting the influence of working in structured, high‐risk environments (Slovic 2013). However, the pattern persisted across both groups, underscoring the inherent difficulty of evaluating abstract outcomes under uncertainty.

These findings align with research on risk perception (Rohrmann and Renn 2000) and prior work (Pascarella et al. 2021) criticizing healthcare risk matrices for overweighting consequences relative to probabilities. This highlights the importance of explicitly addressing uncertainty in professional risk assessments, particularly in healthcare contexts and critical infrastructure.

6.9. Hypothesis Group 9: Differences in Estimation Uncertainty Between Occupational Groups of Healthcare Professionals

Medical professionals consistently reported higher levels of uncertainty across all dimensions, whereas administrative and management staff provided more stable assessments. This suggests that the perception and handling of uncertainty are shaped by role‐specific demands and exposure to complex decision environments. In clinical practice, professionals often face unpredictable outcomes and limited information, which may heighten awareness of uncertainty and promote more cautious estimations. In contrast, non‐clinical roles often operate within structured procedural frameworks that reduce ambiguity and support consistency. These findings align with research showing that frequent exposure to high‐risk or ambiguous situations increases the recognition and expression of uncertainty in judgment (Lipshitz and Strauss 1997). The observed variation highlights the relevance of occupational background when designing and interpreting risk assessment tools.

6.10. Hypothesis Group 10: Differences in Estimation Uncertainty Between Healthcare Professionals Based on Critical Infrastructure Experience

The results of hypothesis group 10 show that healthcare professionals with experience in critical infrastructure contexts differ significantly from those without such experience in how they express estimation uncertainty. Those with infrastructure responsibilities reported lower uncertainty when evaluating impact and potential but higher uncertainty in probability estimates for both risks and opportunities.

This pattern suggests that critical infrastructure experience may strengthen confidence in assessing consequences, possibly due to scenario‐based expertise and familiarity with risk control mechanisms. In contrast, elevated uncertainty in probability assessments may reflect greater awareness of systemic complexity, interdependencies, and unpredictability. Prior research (Aven 2015) highlights that probability estimates in such settings often face data limitations, rare event structures, and systemic uncertainty. These findings are consistent with decision‐making studies showing that professional experience in high‐risk domains can increase judgment accuracy in some dimensions while heightening sensitivity to others (Lipshitz and Strauss 1997).

7. Limitations

This study provides novel insights into the use of graphical risk assessments over traditional methods, but some limitations must be acknowledged. The graphical method allowed respondents to express impreciseness via range‐based estimates, yet it did not differentiate between types or causes of the expert's vagueness. It remains unclear whether broader ranges reflect inherent variability (aleatory uncertainty) or insufficient knowledge (epistemic uncertainty), a distinction that may influence interpretations. Additionally, the study compared two matrix types but did not include repeated assessments using the same matrix. Some observed variation may therefore stem from intra‐individual inconsistencies rather than method effects. Future studies incorporating repeated measures could help disentangle these factors. Although participants were trained professionals familiar with the matrix, scale descriptors, such as “likely” or “high impact,” remain open to subjective interpretation, possibly affecting consistency; we did, however, provide descriptions for all categories to help subjects interpret them. Moreover, the study focused on a specific healthcare context and selected occupational groups, which may limit generalizability to other domains. These considerations do not diminish the value of the findings but help contextualize them. Addressing these limitations in future work could strengthen the robustness of conclusions and inform methodological refinements, such as the use of confidence ratings or metadata on information certainty.

An additional limitation concerns the potential semantic ambiguity of verbal categories used in traditional risk matrices. Although we provided clearly defined descriptors for all scale categories and used an institutionally standardized matrix, previous research has shown that terms such as “high” or “very unlikely” are subject to varied interpretations. Even among experienced professionals, individual differences in how these terms are understood can influence ratings. We aimed to reduce this variability by using a matrix familiar to all participants and by relying on a professional population regularly engaged in structured risk assessment. Nonetheless, the potential for residual semantic variation cannot be fully excluded and may have contributed to the observed differences.

Furthermore, the study's design did not include repeated use of the same matrix to assess intra‐method consistency. This omission limits the extent to which the observed differences between the traditional and graphical assessment methods can be attributed exclusively to their design features. It is possible that some of the differences reflect general temporal variability in expert judgment rather than method effects per se. The study did not include repeated assessments of each method, as this was deemed impractical due to the length and complexity of the survey. In the applied context of healthcare professionals with limited time resources, an expanded design was considered infeasible (by the participants). We therefore opted for a paired comparison that allowed the evaluation of both methods within a manageable timeframe while minimizing dropout and response burden. Nonetheless, we acknowledge that future studies with repeated applications of each assessment format would be valuable to better distinguish between method‐specific effects and the general variability of risk judgments. Building on this, it should also be noted that the absence of retest conditions within the same evaluation format prevents a clear separation of method‐specific differences from intra‐individual variation over time. Although the chosen study design allows for a robust direct comparison between the two methods, some of the observed effects may partially reflect general variability in professional risk perception. Future studies, including within‐method retesting, could help to more clearly isolate design‐related effects from natural fluctuations in expert assessments. Given the applied context and limited resources, the decision to adopt a two‐method crossover design prioritized feasibility and minimized dropout while still enabling a valid comparative analysis. Moreover, although the graphical method enables intuitive communication of uncertainty ranges, it does not provide insight into the underlying causes of these ranges, such as conflicting expert judgments, limited empirical evidence, or divergent assumptions. Such a root cause analysis about the uncertainty was outside the scope of this study, because the graphical method's original purpose (in the literature) was to let people indicate that their assessments are subjectively vague, not also explain their reasons. As highlighted by Goerlandt and Reniers (2016), more advanced methods exist that account for the strength and provenance of supporting evidence. However, these approaches often require greater methodological effort and may be less accessible in interdisciplinary practice settings. The method used here was selected for its balance between conceptual clarity and practical applicability. Nonetheless, the lack of explicit representation of the basis for uncertainty remains a conceptual limitation that warrants further exploration. Indeed, the challenge of “dealing” with uncertainty subsequently in making risk‐driven decisions certainly calls for an underlying account of the source of uncertainty. This problem comes after the initial risk assessment and is a highly relevant open question but is out of this work's scope. Finally, the broader critique of risk matrices as analytical tools remains relevant. Prior work has pointed out structural and conceptual flaws in assigning probability and impact scores or deriving aggregate risk levels, even in qualitative frameworks (Flage and Røed 2012; Goerlandt and Reniers 2016). Our aim is not to defend risk matrices but to investigate how their design affects assessments, given their widespread practical use. Future research is needed to explore not only alternative visual formats but also fundamentally different approaches to representing and communicating uncertainty.

8. Conclusion

This study aimed to examine an alternative graphical method for risk and opportunity assessment in the healthcare sector and to compare it with the traditional assessment approach. The findings reveal significant differences between the two methods, with consistently high inconsistency regardless of occupational group or connection to critical infrastructure. Although the divergence between methods remains stable across subgroups, the perceived uncertainty varies depending on professional background and the presence or absence of a connection to critical infrastructure. Notably, the graphical method may support respondents in reflecting more openly on uncertainty, particularly regarding the impact dimension of risks. These findings highlight the potential of the graphical approach to enhance the quality of risk assessments by making implicit uncertainty more explicit. Additionally, combining traditional and graphical assessment methods may offer further advantages by fostering a more comprehensive understanding of complex risks and opportunities. This may prove particularly valuable in the context of complex, long‐term projects within critical infrastructure sectors, where uncertainty is inherently elevated due to dynamic environments, institutional constraints, and systemic complexity.

The graphical risk approach contributes to a more nuanced understanding of expert assessments and supports informed decision‐making in high‐stakes sectors such as healthcare. In this regard, the ability to not only capture an evaluation but also to express its associated uncertainty could represent a step toward more realistic and reflective risk assessments. Notably, the explicit consideration of uncertainty in individual assessments remains largely underexplored. Our findings suggest that integrating uncertainty systematically into risk assessment practices is feasible and may offer meaningful advantages. The proposed graphical method offers a practical and scalable tool that can help close this gap and foster more transparent, informed, and resilient decision‐making, particularly in the face of increasing complexity within critical infrastructure systems. These conclusions are in‐line with internationally recognized standards that emphasize the consideration of uncertainty as a core element of effective and targeted risk and business continuity management. Furthermore, the results respond directly to recent regulatory developments in the European Union, which require critical infrastructure operators in all member states to conduct risk assessments that account for uncertainty, interdependencies, and evolving threat scenarios. Integrating this dimension into practical assessment tools may support improved governance and more resilient planning strategies, especially within complex, high‐stakes environments such as healthcare systems and other essential service sectors.

Kutej, A. , Rass, S. , & and Alexandrowicz, R. W. (2025). Comparing Traditional and Graphical Risk Matrices: A Case Study in Healthcare. Risk Analysis, 45, 11111. 10.1111/risa.70104

Data Availability Statement

All relevant data files underlying the findings reported in this article are openly available in the Open Science Framework (OSF) repository at: https://doi.org/10.17605/OSF.IO/GTP5A.