Examining water risk perception and evaluation in the corporate and financial sector: a mixed methods study in Ontario, Canada

As primary users of a socially, economically, and environmentally significant yet increasingly stressed resource like water, the corporate and financial sectors have an important role in sustainable water management. However, extant literature reveals a gap in the empirical assessment of water risk perception and its influence on water risk evaluation and decision-making in the corporate and financial sectors. Our explanatory sequential mixed methods study examined the relationship between water risk perception and risk evaluation (risk ratings), addressing these gaps. We employed a cross-sectional survey (N = 25) followed by semi-structured interviews (N = 22), with a purposive expert sample of analysts, practitioners, and decision-makers in the corporate and financial sector in Ontario, Canada. Our study finds multi-dimensional risk perception factors, including knowledge, professional experience, perceived controllability, values, trust, location, and gender, that influence water risk ratings and vary with the type of risk. Moreover, the in-depth follow-up interviews reveal multiple drivers of different risk ratings, such as proximity bias, sector differences, trust in various institutions, as well as the influence of tacit knowledge, exposure, the role of regulations, media, and financial materiality. Our study empirically concludes that the water risk perception of analysts, practitioners, and decision-makers in the corporate and financial sectors is highly nuanced and impacts the evaluation of different water risks, and should be systematically integrated into risk assessment and decision-making frameworks. Our study advances knowledge in the fields of risk analysis and sustainable water management and contributes by empirically examining and explaining the complex and underexplored relationship between water risk perception factors and evaluation using novel interdisciplinary Risk Theory and mixed methods approaches. Finally, the study’s findings can help integrate sector and location-specific preferences and priorities with analytical data to design contextually-attuned decision support tools for sustainable water management strategies, policies, and practices.


Introduction
Amidst growing risks to water resources across the globe, Sustainable Development Goal 6 (SDG 6) emphasizes sustainable water management to perpetually safeguard water availability, quality, ecosystems, and access for current and future generations (United Nations 2018, Sandhu et al 2021).Water security and sustainable water management are inextricably linked to assessing and managing water risks by all water-using sectors to reconcile competing values, interests, and highly variable water resource supply (Di Baldassarre et al 2019).Nonetheless, despite the significance of water security and its interconnection with sanitation, health, food and energy security, economic productivity, and environment, the progress on SDG 6 has mainly been unsatisfactory 1.1.Risk perception of experts and practitioners Risk is an objective and subjective construct that is perceived and interpreted by human actors and is further shaped by the object of risk (source of risk, e.g., water scarcity), perceived impact of risk (what is at risk, e.g., reputation, profits, legitimacy, and health), as well as the affective, trust-based, and socio-cultural characteristics of the individual risk perceiver (Dobbie and Brown 2014, Klinke and Renn 2019, Siegrist and Árvai 2020).Decision-making for complex risks entails problem framing, risk assessment, followed by risk evaluation/ characterization, where analysts and decision-makers review technical evidence (objective component) and assign a priority based on the acceptability or tolerability of risk (value-based component) (Klinke andRenn 2012, 2019).Thus, contrary to dominant psychometric paradigms, experts are not a value-free homogenous group but have nuanced cognitive (analytical) and affective (connected to past experiences, feelings, and emotions) risk perception mechanisms that influence risk evaluation and decision-making (Slovic 1999, Sjöberg 2002, Siegrist and Árvai 2020).
It is important to note that 'experts' not only include academic experts but also practicing experts, i.e., analysts, practitioners, and decision-makers in corporate organizations, government, or civil society who undertake risk assessments to inform management and decision-making related to the risk problem (Sjöberg 2002, Dobbie andBrown 2014).Thus, given multiple decision points, the practitioner/analyst/ decision-maker's risk perception informs value-based judgment and prioritization regarding different risks, playing a critical role in risk analysis and management strategies (Dobbie and Brown 2014).Risk problems, such as water, are multi-faceted, complex, ambiguous, and uncertain, with little or excessive conflicting knowledge (Klinke and Renn 2019).Thus, practitioners' perception and prioritization of various water risks tend to vary based on their sector, roles, experience, and awareness of impacts (Dobbie andBrown 2014, Siegrist andÁrvai 2020).
As acknowledged by risk evaluation, management, and governance frameworks, an essential component of comprehensive risk analysis is the examination of the role of water risk perception of practitioners and decisionmakers in different water-using sectors and the underlying cognitive and affective factors (Dobbie andBrown 2014, Renn et al 2020).However, extant research on attitudes and perception of water risks has mainly focused on the lay public using quantitative methods like surveys (Sjöberg 2002, Dobbie and Brown 2014, Mumbi and Watanabe 2020, Sandhu et al 2021).Methodologically, mixed methods are an emerging method used to obtain a thorough understanding and explanation of factors underlying risk perception and their relationship with risk evaluation that are underexplored in extant literature (Quinn et al 2019, Siegrist 2019, Siegrist and Árvai 2020).Therefore, our research focuses on the novel application of mixed methods to investigate the perception of biophysical and social water risks by corporate and financial practitioners and decision-makers.

Water risk perception in the corporate and financial sector (CFS) in Ontario
The economy of Ontario in Canada is driven by its large manufacturing and financial sector.Surrounded by the Great Lakes, Ontario is often considered an ideal location for water-reliant production, agriculture, and investment (Sandhu et al 2020).Despite the perception of abundant freshwater resources, Ontario is rife with water issues, including decreasing flows, quality issues, depleted and /or contaminated groundwater sources, regulatory uncertainty, legacy water access issues in indigenous communities, and controversies/conflicts among competing water user groups, especially the corporate sector (Johns 2017, Bonsal et al 2019, Sandhu et al 2023).Moreover, the role of non-state expert and practitioner risk perception in assessing and managing different water risks largely remains underexplored in Ontario (Sandhu et al 2021(Sandhu et al , 2023)).Therefore, Ontario is a contemporary and instructive case for empirically examining the underlying facets of water risk perception of practitioners and decision-makers and how biophysical and social water risks are assessed, ranked/prioritized, and integrated into decision-making (Sandhu et al 2021).
Consequently, our study advances knowledge in risk analysis and sustainable water management.It makes an original and novel contribution by empirically examining and explaining the complex and underexplored construct of water risk perception and its relationship with risk evaluation.Secondly, the study exemplifies the novel application of Risk Theory and explanatory sequential mixed methods approaches for water risks.By analyzing the multidimensional construct of water risk perception and its influence on water risk evaluation and decision-making, multi-sector policies and practices can be designed for sustainable water management.Moreover, this study's research design and analytical procedures can be tested and applied to other regions in future studies beyond Ontario (Creswell and Creswell 2018).

Research objective and questions
Extant literature reveals gaps in examining the role of water risk perception in evaluating and managing water risks in the CFS in Ontario.Addressing these gaps and using the case of Ontario, Canada, our explanatory sequential mixed methods study aims to empirically examine water risk perceptions and priorities of practitioners and decision-makers in the CFS, as well as the relationship of water risk perception with water risk evaluation and decision-making.
RQ 1: What are the factors underpinning the water risk perception of practitioners and decision-makers in the CFS in Ontario?
RQ 2: How does water risk perception relate to water risk evaluation in the CFS?
2. Hypotheses and theoretical framework for water risk perception and evaluation in the CFS For examining a complex, uncertain, and ambiguous risk problem like water and water risk perception, a single siloed theory, method, or disciplinary paradigm may be insufficient (Dobbie and Brown 2014, Klinke and Renn 2019, Kasperson et al 2022).Extant literature highlights the significance of using interdisciplinary approaches to tease out complexities of risk perception and evaluation by integrating multiple physical and social theories and methods across disciplines (CohenMiller and Pate 2019, Dobbie and Brown 2014, Klinke andRenn 2019, Siegrist andÁrvai 2020).Risk Theory is a broad, interdisciplinary, inclusive, and all-encompassing theory that tethers multiple approaches and concepts from disciplines like engineering, psychology, sociology, technology, and political science to get a panoramic systems-based perspective on risk and risk perception (Roeser et al 2012, Renn et al 2020).Thus, given the theoretical necessity of interdisciplinary and integrated approaches for examining complex constructs like water risk perception, we chose Risk Theory over other siloed theories to address our research questions.
Using Risk Theory, we can integrate theoretical constructs and hypotheses from psychrometric and sociological paradigms to develop our comprehensive interdisciplinary framework for water risk perception and evaluation (CohenMiller and Pate 2019, Dobbie andBrown 2014, Klinke andRenn 2019).For instance, aligned with the psychometric paradigm, risk perception and evaluation can vary with the type and nature of risk and are contingent on the characteristics of hazards (Klinke and Renn 2019).Moreover, sociological theories like Cultural, Value-Belief-Norm, and Relational Theory state that risk perception is contingent on the risk perceiver (Dobbie andBrown 2014, Siegrist andÁrvai 2020).Integrating both approaches, we included physical, cognitive, affective, and socio-cultural factors in our hypotheses and framework.Therefore, Risk Theory provides an apt interdisciplinary foundation to extend and apply the analytical-normative framework of risk governance to holistically examine water risk perception factors as well as unearth the complex relationship of these factors with water risk evaluation and management (Roeser et al 2012, Renn et al 2020, Kasperson et al 2022).

Factors underlying water risk perception for risk evaluation and management
Based on Risk Theory and its underlying physical, psychometric, sociological, cultural, and organizational theories, cognitive, affective, and socio-cultural factors of risk perception are expected to act like filters shaping water risk perception of CFS practitioners and influencing risk evaluation (dependent variable), i.e., ratings or priorities assigned to different biophysical and social water risks (Sjöberg 2002, Dobbie and Brown 2014, Vasvári 2015).Water risks relevant from a corporate and financial perspective include water quantity (droughts, groundwater depletion, or reduced flows), quality (contamination), regulatory stringency and uncertainty, location-specific water user conflicts (competing users), source sensitivity (groundwater versus surface water), sector-specific risks, and public and media scrutiny (Money 2014b, Sandhu et al 2023).
The factors and corresponding hypotheses are based on general risk perception theories and governance frameworks across multiple disciplines tethered under Risk Theory that we adapted for the risk domain of water in the CFS to develop our study's integrated theoretical framework (CohenMiller and Pate 2019, Kasperson et al 2022).Given the research objective, all hypotheses relating risk perception factors (independent variables) to water risk evaluation are associative (explanatory), not causative or predictive.In the case of directional hypotheses, positive association implies that an increase in the factor tends to increase the risk rating, whereas negative association indicates a decrease in ratings.
2.1.1.Nature of water risk 'Nature of risk', drawn from the psychometric paradigm, focuses on the risk object, i.e., water risks (Mumbi and Watanabe 2020).This factor captures the physical conceptualization of different types of water risks, their likelihood, drivers, prioritization, and extent of integration in decision-making that tend to shape risk perception and evaluation (Dobbie andBrown 2014, Mumbi andWatanabe 2020).Consequently, our first hypothesis is: H1 Nature of Risk : There are differences in practitioners' conceptualization and prioritization of biophysical and social water risks (Dobbie andBrown 2014, Klinke andRenn 2019).

Water attitudes
Drawing from sociological theories, attitudes are psychological tendencies about a phenomenon that individuals develop based on their experiences, impacting their evaluation (positive or negative) as well as behaviour and actions towards that phenomenon (Eagly and Chaiken 2007).Attitudes are domain-specific, i.e., not homogenous across all risks, whereas the Theory of Reasoned Action and the Theory of Planned Behaviour can help explain behavioral and perception differences (Weber et al 2002).Attitudes towards different risks are expressed through multiple variables, including scope, controllability, familiarity (exposure or experience with the issue), concern (dread or worry), urgency, awareness of the negative impact, benefits, and perceived equity of the risk (McDaniels et al 1997, Slovic 1999, Mumbi and Watanabe 2020).For this study, we adapted these variables for water and developed the following hypotheses (McDaniels et al 1997, Slovic 1999, Dobbie and Brown 2014, Mumbi and Watanabe 2020).
H2 Attitude Scope : Perceived scope or extent, i.e., the area and people impacted by the specific water issue, is expected to be positively associated with one's risk perception and, hence, the rating of that water risk.The higher the perceived scope of impact, the higher the risk perception and rating of risk.
H3 Attitude Controllability : Controllability is expected to be negatively related to one's risk perception and hence risk rating.
H4 Concern : Concern is expected to be a multi-faceted construct encompassing dread, worry, perceived unfairness or inequity related to costs and benefits distribution water of risks, non-substitutability of benefits to offset costs and risk, as well as perceived urgency.
H5 Exposure : Direct exposure to water issues is expected to be positively related to risk perception, concern, protective behaviour, and risk ratings (Thistlethwaite et al 2018).

Water-related knowledge
Drawing from the relational theory of risk perception, knowledge is a critical cognitive factor related to core mental processes, where information related to the risk problem is processed to generate risk perception (Dobbie andBrown 2014, Siegrist andÁrvai 2020).CFS practitioners have individual and collective knowledge as well as interdependent overt and tacit knowledge, which create individual-level differences in perceptions of the same risk.While overt knowledge is connected to formal education and training, tacit or interpretative knowledge is the implicit understanding of phenomena that are gained subconsciously through experience, values, beliefs, and context (Wolfe 2009, Klinke andRenn 2019).Tacit knowledge can potentially be impacted by water-related values, attitudes, and previous exposure to issues (Dobbie and Brown 2014).Integrating the construct of knowledge in our study, water is an interdisciplinary risk problem, and planners, accountants, economists, engineers, ecologists, or management professionals in the same sector may prioritize water risks differently (Dobbie and Brown 2014).Moreover, while lack of knowledge increases uncertainty, competing claims or multiple interpretations of complex risk phenomena like water may increase risk perception (Weber 2001, Aven andRenn 2019).Thus, knowledge and awareness of the complexities of water issues, their impact, and direct exposure tend to increase risk perception (McDaniels et al 1997, Klinke andRenn 2019).
H6 Knowledge : Knowledge encompassing own expertise, awareness of adverse impacts, and experience is expected to be positively associated with water risk ratings.

Water values and beliefs
Drawing from the sociological theories like Value-Belief-Norm Theory and Cultural Theory, values and beliefs are expressions of worldviews and overarching sets of goals and principles (distinct but related to attitudes) that guide an individual's behaviour and is posited as an explanatory factor for risk perception, especially in conjunction with knowledge (Slimak and Dietz 2006, Schwartz et al 2012, Dobbie and Brown 2014).Domainspecific values, primarily environmental values, have been studied to understand the environmental behaviour of different stakeholders (Slimak and Dietz 2006, Krewski et al 2008, Bøhlerengen and Wiium 2022).For such risk domains, values like 'Biospheric' (environmental concern), 'Altruistic' (concern for others), 'Egoistic' (concern for self, assessing costs and benefits), 'Technological Optimism' (optimism in technology to address issues) have been identified (Krewski et al 2008, Bouman et al 2018).Moreover, biospheric and altruistic values are collectively considered sustainability-centric values (sustaino-centric), and economic benefits-centric values have also been identified in sustainability management, where perceived financial benefits or impacts tend to shape risk acceptability and perception (Gladwin et al 1995).Adapting these broader environmental values for water, we posit the following hypotheses for risk evaluation: H7 Sustainocentric Value : The higher the sustainability-centric values of an individual, the higher the risk ratings due to higher concern for the environment and others, including future generations.
H8 Egocentric Value : The higher the egocentric values, the lower the perceived risk and ratings for water H9 Economic Benefits Value : Economic benefits-centric values are related to risk ratings but are expected to vary across risk types.
H10 Technological Optimism Value : The higher the optimism in technological solutions, the lower the risk ratings.

Trust in various institutions
Drawing from political science and governance paradigms, trust is an increasingly important yet complex explanatory variable of risk perception and evaluation, connected to heuristics, where in case of limited or competing knowledge, individuals may rely on trust to aid decision-making hence impacting risk acceptability (Siegrist 2019, Mumbi andWatanabe 2020).Trust can be multi-dimensional and variable, where experiences with an institution and qualities like communication, neutral interests, transparency, and past positive performance can help build (competence-based) trust and confidence.Moreover, social trust can be based on value similarities and group membership (industry) (Slovic 1999, Dobbie and Brown 2014, Siegrist 2019).Therefore, trust is a relevant factor for water risk perception and evaluation.H11 Trust : The higher the trust (competence-based) in the government or in an industry (social trust) to assess and manage water risk, the lower the risk perception and risk ratings, i.e., negative association.

Sociocultural demographic characteristics of CFS experts and practitioners
Drawing on cultural and relational theories of risk perception, extant literature finds a highly variable impact of sociocultural demographic characteristics like gender, ethnicity, location, professional role, sector, etc, on risk perception that is typically controlled instead of being explored as a part of a theory (Krewski et al 2006, Slimak and Dietz 2006, Siegrist and Árvai 2020).For instance, some studies conclude that women have higher risk perception in some risk domains, but some studies report no effect (Dupont et al 2014, Mumbi andWatanabe 2020).Slovic (1999) attributed the higher concern of women about human health, environment, well-being, and safety to the social norm of being entrusted with nurturing and maintaining life.Culture is an interesting factor defined as a dynamic set of shared or group beliefs, norms, meanings, customs, and values that are acquired by an individual (Weber and Hsee 1998).Therefore, in our study, cultural importance refers to practitioners' shared beliefs, attitudes, or trust related to water bodies, expected to influence risk perception (Dupont et al 2014, Siegrist 2019).Other factors include the sector of water use, academic discipline, and location (Dobbie and Brown 2014).
A location/proximity variable is useful in examining proximity bias, where, interestingly, an individual may discount the occurrence of water issues in their own sub-watershed due to higher perceived control, hence impacting concern, confidence, risk perception, and assessment (Weber et al 2002, Krewski et al 2008, Money 2014b, Quinn et al 2019).Instead of controlling these sociocultural demographic variables, we empirically examined them as part of our theoretical framework, included as explanatory variables (Mumbi andWatanabe 2020, Siegrist andÁrvai 2020) H12 Gender : Women are expected to have higher water risk ratings than men.H13 Cultural Importance : The cultural importance of water is expected to be correlated to trust and confidence.H14 Sector : The sector of water use is expected to influence water risk ratings, where water-intensive sectors rate/prioritize water risks higher than others.
H15 Location : A proximity bias is expected, where an individual may have lower concern and risk perception of their local sub-watershed than the province (as a whole).

Theoretical framework
Based on the investigated risk perception factors and hypotheses (section 2.1) drawn from underlying interdisciplinary paradigms of the Risk Theory that we adapted for water and aligned with the integrated risk analysis stage of the normative-analytical risk governance model of Klinke andRenn (2012, 2019), we developed the study's theoretical framework as depicted in figure 1.These cognitive, affective, social-cultural, and spatial (proximity) factors are posited to shape water risk perception, influence risk evaluation, and, eventually, risk management and decision-making for water.The risk governance model of Klinke andRenn (2012, 2019) is a generic model for all risk domains, encompassing all stages of risk management, including risk pre-estimation, interdisciplinary risk estimation, integrated risk analysis (perception and evaluation), management, and communication.We tested this model by applying it to water risks in the CFS.Then, we built it further by expanding the risk evaluation stage to purposefully include the integrated water risk perception factors drawn from interdisciplinary theories.

Methods and data
Recent literature advocates using mixed methods to address multi-faceted research problems like water risks to ensure quantitative robustness and qualitative depth in tandem (Creswell and Creswell 2018, Quinn et al 2019, Mooney et al 2020).Using a single method will not suffice to examine and understand complex risk perception factors holistically, and there is a paucity of studies employing mixed methods for investigating water risk perception and evaluation (Di Baldassarre et al 2021, Quandt 2022, Rangecroft et al 2021, 2022).Therefore, demonstrating the novel application of mixed methods, we employed an Explanatory Sequential Mixed Methods research design, as depicted in figure 2. We combined the quantitative robustness of cross-sectional surveys and the explanatory depth of follow-up qualitative interviews to address our research questions.
As human participants were involved, the study was reviewed by the University of Waterloo Research Ethics Board (REB# 44065), ensuring ethical recruitment, informed consent, data collection, anonymization, and management.The quantitative stage examined the underlying factors for water risk perception and their relationship with evaluation (captured as risk rating or weights for different water risks).We used a questionnaire-based cross-sectional online survey designed in Qualtrics for data collection (Quinn et al 2019, Renn et al 2020).The follow-up qualitative stage with in-depth one-on-one semi-structured interviews was designed to complement the survey.It was conducted with the same sample of participants as the survey (Creswell and Creswell 2018).

Study participants and sampling strategy
The sociopolitical landscape for sustainable water management is very diverse with multi-level systems, i.e., a multitude of stakeholder groups including public (state) actors, policy makers, private actors, and civil society, along with their diverse perceptions, values, and interests (Klinke andRenn 2012, Johns 2017).However, extant research has primarily focused on the public sector, policy makers, academia, and lay public (Johns 2017, Harris-Lovett et al 2019, Klinke and Renn 2019, Quinn et al 2019, Mooney et al 2020).As discussed before, private actors, including the CFS, are essential stakeholders in sustainable water management but have been an underexplored population in the water risk assessment and perception research (Money 2014a, 2014b, Christ and Burritt 2018, Hogeboom et al 2018).
Addressing this gap, private sector practitioners and experts, including analysts, managers, and decisionmakers from businesses in water-reliant industrial sectors (e.g., food and beverage, aggregate mining, chemical manufacturing, other manufacturing, power production, agriculture, environmental consulting, and research services), and financial sector (e.g., banks, investors, insurance companies etc.), who have expertise in environmental/sustainability risk assessment in Ontario, were chosen as the study's population.An expertbased purposive sampling strategy ensured that the attitudes and perceptions of analysts and decision-makers in different sectors are considered (Mooney et al 2020).Informed by extant research, purposive sampling is strategic and ensures the representativeness and relevance of the selected sample to the research objective geared toward in-depth context-specific insights (Palinkas et al 2015, Mooney et al 2020).
We identified 252 potential participants after research ethics clearance based on publicly available professional profiles on LinkedIn, industry associations, and organizational websites.We used the job title, sector, and location filters to ensure representativeness.The following purposive recruitment criteria were used to screen and select potential participants: (i) Professional experience or expertise (based on their roles/work experience) in sustainability, environmental, water assessment, reporting, management, and decision-making (ii) Job titles including 'sustainability', 'environment', 'natural resource', 'water', 'ESG', 'water policy' AND 'analyst ', 'technician', 'engineer', 'manager', 'director', 'lead', 'specialist', 'advisor', 'associate', 'vice president', 'president', 'consultant' (iii) Current or past work location in Ontario (Northern Ontario, Southern Ontario).
We sent invitation emails along with the information and consent letter to the sample.The information/ consent letter provided an overview of the study, objectives, significance, and time commitments.We also provided information about the anonymity and confidential procedure, timelines, scholarly/scientific benefits, consent to record, and the use of quotes.The online survey was available from April 25 to August 10, 2022.Twenty-six participants provided consent to participate, and 25 participants completed the online survey.From the recruited sample, 22 participants participated in the follow-up online interviews between October 1 and December 15, 2022.Demographic data of the survey and interview participants are provided in table 1.
Comment on the Adequacy of Sample Size: Studies that employ expert surveys and interviews tend to have a highly specific and small sample (McDaniels et al 1997, Weber 2001, Mooney et al 2020).Arguments about the adequate sample size to ensure representativeness using power analysis, minimum variable-to-case ratios, and restrictions on parametric tests are rife in the literature (Norman 2010, Jenkins andQuintana-Ascencio 2020).Nonetheless, the representativeness of a population can be ensured using non-probability based intentional purposive sampling strategies (Palinkas et al 2015, Mooney et al 2020).Moreover, parametric tests do not strictly prescribe a minimum sample, and the suggested variable-to-case ratios are rules of thumb varying considerably across disciplines and study designs (Norman 2010, Jenkins andQuintana-Ascencio 2020).The emphasis on large samples and power analysis has been criticized for being overly prescriptive, overlooking particularity, i.e., gaining contextual insights of highly specific research populations like experts (Jenkins and Quintana-Ascencio 2020).
Statistically, a smaller sample size requires larger effects to achieve statistical significance, and thus Type II errors or false negatives might be a concern (Norman 2010).Nonetheless, in explanatory sequential mixed methods research, the survey's findings are validated and explained in-depth during the follow-up interviews, enhancing the overall research validity and reliability (Ivankova et al 2006, Creswell andCreswell 2018).Moreover, parametric tests, including multiple linear regression, are robust for sample sizes as small as 15, provided the basic assumptions of multiple linear regression analyses, i.e., linearity, low multi-collinearity, normality of residuals, and homoscedasticity are met (Norman 2010).Simulations that matched regression models to data found results to be stable for sample sizes around 25 across variable effect sizes (Jenkins and Quintana-Ascencio 2020).Thus, in our study with 25 experts, not only are the explanatory variables based on the extensive literature on risk perception but also the follow-up interviews help validate significant and nonsignificant explanatory variables to address Type II errors (Molina-Azorín and López-Gamero 2016, Creswell and Creswell 2018).

Survey questionnaire design
Survey items aligned with variables and measures were drawn from extant risk perception literature further adapted for water risks and the context of Ontario (Slimak and Dietz 2006, Krewski et al 2006, 2008, Dupont et al 2014, Bouman et al 2018, Robinson 2018, Thistlethwaite et al 2018, Mumbi and Watanabe 2020, Grima et al 2021).Operationalizing our theoretical framework, water risk perception factors (independent variables) to be examined included Nature of Risk (different water issues, likelihood, drivers, extent of integration), Attitudes (confidence, scope, controllability, equity, impact awareness, benefits, previous exposure to water issues, overall concern, and urgency), Knowledge (assessment of experience, knowledge in different sectors, source of knowledge), Values and beliefs (biospheric, altruistic, egoistic, technological optimism), Trust (in institutions to manage water risks), and Sociocultural demographic characteristics (cultural importance of water, sector, discipline, professional role, gender, ethnicity, location).Sociocultural demographic questions were selfreported, and the remaining questions were assessed on a 7-point Likert scale.Water risk evaluation (dependent variable) was measured by assigning a rating (priority) to seven types of water risks (figure 1) using a 1 to 7 continuous scale based on the perceived importance of the risk to business and investment operations, policies, or decisions (McDaniels et al 1997, Weber 2001, Dobbie and Brown 2014).
Further details on the survey items and scales are provided in the supplementary material (Appendix A).Given the length of the survey, it was not timed and could be filled out by the participants in multiple sittings at their convenience.Moreover, the participants could skip any question they did not feel comfortable answering.

Statistical analysis of stage 1 survey
We used IBM SPSS, V.28 to analyze the survey dataset containing numerically coded values and labels (anchors of rating scale) for each item.We coded the categories manually for text entry options, e.g., Discipline of education, Professional role, Location, i.e., sub-watershed of residence/work.Categories of a few variables were combined into a broader 'others' category to ensure at least two cases for each category.A new variable, 'Location Conflict Rating' was created from 'Location' to link the actual water conflict risk of the participant's sub-watershed based on the technical water risk assessment by Sandhu et al (2023) to the perceived risk ratings.The location conflict risk (1: very low to 5: very high) for sub-watersheds is based on the density of water-taking permits, drought potential, high growth regions, the prevalence of drinking water advisories, public and civil society focus, and media attention (Sandhu et al 2023).For the regression analysis, we created dummy variables for non-ordered categorical variables, i.e., gender, sector, ethnicity, and discipline of education.
Based on extant studies, Principal Component Analysis and Varimax/orthogonal rotation were employed as the factor extraction and rotation method, respectively (McDaniels et al 1997, Robinson 2018, Grima et al 2021).
We followed the thresholds of Kaiser-Meyer-Olkin (KMO) >= 0.5 and Bartlett's Test of Sphericity, i.e., p <= 0.05 as measures of sample size adequacy and robustness of analysis (Mumbi andWatanabe 2020, Grima et al 2021).Items with factor loadings >= 0.40 in the rotated matrix were retained, and the number of factors/ components was based on eigenvalues > 1 (Slimak andDietz 2006, Boateng et al 2018).Cross loadings of items, if found, were assessed case by case, leading to either removal or being assigned to one factor based on the theory further validated by reliability tests (DiStefano et al 2009, Schwartz et al 2012).
The reliability of each component/construct post-factor analysis was determined using Cronbach alpha (α) (acceptable >= 0.7) (Lam 2012, Robinson 2018).Due to the small sample size and the complexity of a few constructs, additional criteria of composite reliability (CR) (acceptable > 0.6) and Average Variance Explained (AVE) (acceptable > 0.4) were used to ensure reliability and convergent validity (Fornell and Larcker 1981, Lam 2012, Bouman et al 2018).The names and definitions of constructs were assigned based on the literature and theory (Robinson 2018, Grima et al 2021).Recent literature on behavioral research suggests that the robustness of sum or mean item scores post-factor analysis is comparable to more refined weighted factor score estimates with the additional advantage of simplicity and consistency across samples (Boateng et al 2018, Robinson 2018, Widaman and Revelle 2022).Thus, upon identification of constructs, the average raw scores of constituent items were used.Since average scores were used, the original scale range (1-7) was retained for the new constructs.Items with negative signs for factor loadings in a construct were reverse-coded and then included in the mean scores before being used in statistical tests (Slimak and Dietz 2006, Bouman et al 2018, Robinson 2018).

Statistical tests to examine the relationship between water risk perception and evaluation
After factor analyses, we used Multiple Linear Regression (Ordinary Least Squares estimation method) to analyze and explain the relationship between each water risk type (dependent variable) and hypothesized explanatory risk perception factors measured on the Likert scale (McDaniels et al 1997, Slimak and Dietz 2006, Shmueli 2010, Mumbi and Watanabe 2020).Past studies have also employed parametric tests for Likert scale data that are considered to be continuous with five or more categories (Weber 2001, Slimak and Dietz 2006, Norman 2010, Mumbi and Watanabe 2020, Bøhlerengen and Wiium 2022).Multiple linear regression models were developed for the three water risk ratings, and the models were intended to be explanatory rather than predictive or causal (Weber 2001, Shmueli 2010).
For each model, we used the standard method of simultaneous entry of independent variables (ENTER method), listwise deletion of missing data, and dummy variables for non-ordered categorical data (Mumbi and Watanabe 2020).We calculated standard measures like Variance Inflation Factor (VIF) with a conservative acceptable threshold of < 4 (to minimize multicollinearity), adjusted R 2 , F value significance (two-tailed test with alpha of 0.05), and tested all statistical assumptions post hoc (Slimak and Dietz 2006, Boateng et al 2018, Mumbi and Watanabe 2020).Similar to past perception studies, the number of explanatory variables was relatively high for water risk perception and evaluation (McDaniels et al 1997, Slimak and Dietz 2006, Mumbi and Watanabe 2020).Nonetheless, to diagnose overfitting and ensure apt model selection, in addition to adjusted R 2 , the PRESS (predicted residual sum of squares) statistic with the total sum of squares (SSTO) of the model was calculated and compared (Allen 1971).
Additional multiple linear regression models were developed to test the hypotheses related to the factors of Trust and Confidence (as dependent variables) and identify underlying explanatory factors (Mumbi and Watanabe 2020).Moreover, to test statistically significant mean differences based on gender and sector for variables of interest, we employed independent samples Student's t-test (for two groups for gender) and one-way analysis of variance (ANOVA) test (Mishra et al 2019).We also performed normality (Shapiro-Wilk Test) and homogeneity tests (Levene Test), and the differences were validated in follow-up interviews (Ivankova et al 2006, Creswell andCreswell 2018).All significance tests for regression or other tests, i.e., p < .05,was a two-tailed test with an alpha of 0.05 (Mishra et al 2019).

Stage 1 survey results
4.1.Participant demographics and descriptive statistics (mean scores) for variables of interest A total of 25 (N Survey ) participants completed the survey, and 22 (N Interview ) participated in the follow-up interviews (table 1).For variables of interest, specifically, Trust (degree of trust in different institutions to assess risks, manage or protect water resources), the mean score of all participants was highest for Civil Society Organizations (M = 4.28, SD = 1.14), followed by Government (M = 4.12, SD = 0.73) and least in Private Sector (M = 3.48, SD = 0.87).To explore Proximity Bias, we found the mean score of all participants for Confidence (in sufficient and abundant water resources) was higher for one's own sub-watershed of residence/work (M = 5.21, SD = 1.14) than Confidence for Ontario more broadly (M = 5.13, SD = 0.85).Moreover, the mean score for Concern (about water issues, their impacts, and risks) was higher for Ontario (M = 5.25, SD = 1.11) than concern for one's own sub-watershed of residence or work (M = 4.79, SD = 1.18).Therefore, confirming proximity bias, the study finds potential discounting (on average) in perceived water risks in one's own subwatershed, i.e., less concern and more confidence than in Ontario.While overall Concern for water issues in Ontario is higher than the confidence in water abundance in Ontario, there is more confidence than concern in one's sub-watershed.4.2.Results of factor analysis, final constructs, and mean water risk ratings We found three constructs for the dependent variable from the EFA based on the seven water risk types.Two constructs, Indirect Water Scarcity Risk (i.e., water quality and source type risk) and Social Water Risk (regulatory risk, water user conflict, sector-specific risks, and media/public attention), were found with a cumulative explained variance of 66.75% and no cross-loadings.Direct Water Scarcity Risk (water quantity) was unidimensional based on the literature and excluded from the EFA.For risk perception factors, individual EFAs were performed due to the complexity of the constructs (Grima et al 2021).Two constructs were found for Nature of Risk, i.e., Biophysical Aspects and Social Aspects, with a cumulative explained variance of 60.74% and no cross-loadings.Three constructs were found for Knowledge, i.e., Self, General Issues, and Experts, with a cumulative explained variance of 73.32%.Q15_1_Knowledge_Sustainability Assessment cross-loaded in Knowledge_Self and Knowledge_General Issues, but based on the theory, it was retained in Knowledge_Self.Following Schwartz et al (2012), the assignment to a particular factor was validated by comparing α, CR, and AVE values with higher values resulting in the most appropriate scale.
Two constructs were found for Drivers of Water Issues, i.e., Micro (Consumer) Level and Macro Level, with a cumulative explained variance of 71.30% and no cross-loadings.EFA for Water Values was expected to be complex and iterative, with the final model revealing four constructs, i.e., Sustaino-centric, Economic Benefits centric, Ego-centric, and Technological Optimism, and a cumulative explained variance of 78.93%.Cross-loadings were observed for Q20_7_Efficiency gains can reduce water risks, loading on Economic Benefits centric and Egocentric values.However, based on theory, the factor was assigned to Economic Benefits further validated by comparing α, CR and AVE values.Cross-loadings also appeared for Q20_4_Need to assess both costs and benefits, loading on Ego centric values, Sustaino-centric, and Technological optimism values.The factor was assigned to Egocentric values, and this assignment resulted in the highest α, CR and AVE values.Ego-centric and benefit-centric values were expected to be complex with relatively low α (Bouman et al 2018), and hence we relied on CR > 0.7.
Reputational Impacts and Financial Impacts were two constructs with a cumulative explained variance of 88.40% and no cross-loadings.A separate EFA was performed for Sustainability Impacts with a cumulative explained variance of 86.45% and no-cross loadings.Five constructs resulted for Attitudes, i.e., Scope, Biophysical Controllability, Social Controllability, Overall Concern, and Overall Confidence, with a cumulative explained variance of 73.47%.Cross-loadings were observed for Q7_5_ Controllability_Water user conflicts, loading on Scope and Social Controllability.The factor was assigned to Social Controllability that resulted in the highest α, CR and AVE values.Negative factor loadings for items Q8_Equity of Impact of Water Issues and Q9_Economic benefits offset costs and risks were reverse-coded, alluding to higher concern with higher inequity and nonsubstitutability of water risks.One construct was found for Exposure with a cumulative explained variance of 75.08% and no cross-loadings.
Detailed results of the eight EFAs are provided in Appendix B (Tables B1-B8), which reports the construct names, underlying items, factor (component) loadings, eigenvalues, % of total variance explained, KMO, Bartlett tests, and reliability tests (α, CR, AVE).All factor analyses yielded KMO values > 0.50 and a significant Bartlett's test of sphericity, i.e., p < .05,alluding to the adequacy of the sample and factors.Any item with negative factor loadings was reverse-coded, followed by reliability analysis and averaging for the final construct (Robinson 2018, Grima et al 2021).CR > 0.7 was found for all final constructs, which meets the minimum threshold of 0.6 for reliability (Fornell and Larcker 1981, Lam 2012, Grima et al 2021).Before proceeding with the statistical tests, we created a final set of 23 constructs based on average item scores.For example, Concern was the average of four survey items (Q12_1, Q13, Q8_REV, Q9_REV), where Q8 and Q9 were reverse-coded due to negative factor loadings.
For the dependent variable, based on the EFA, the highest average score was found to be Direct Scarcity Water Risk, i.e., risk of water quantity issues (M = 4.88, SD = 1.68), followed by Indirect Water Scarcity Risk, i.e., scarcity due to degraded water quality and groundwater sources sensitive to contamination (M = 4.83, SD = 1.40), and least Social Water Risk (M = 4.80, SD = 1.19).

Results of the statistical analysis
We developed three multiple linear regression models for water risk ratings, i.e., Direct Water Scarcity Risk, Indirect Water Scarcity Risk, and Social Water Risk (dependent variables).Based on the theoretical framework and EFA results, our study had 21 explanatory variables.To arrive at a relevant sub-set of explanatory variables (risk perception factors) for each model, we compared VIF, Adjusted R 2 , F value, and PRESS statistic.The final models resulted in a significant equation.The Direct Water Scarcity Risk multiple regression model (table 2) indicated that the selected twelve risk perception factors explained 87% of the variance (adjusted R 2 = 0.87, R 2 = 0.94, F(12, 11) = 13.85,p < .001).The factors, along with their unstandardized (B), standard errors, standardized coefficients (β), t values, p values, and VIF, are provided in table 2. Nine out of the twelve explanatory risk perception factors, including Location Conflict Rating, Education Level (College), Education Level (Bachelor, University), Attitude_Scope, Knowledge_Experts, Gender (Woman), Attitude_Bipohysical Controllability, Values_Egoistic, and Values_Technological Optimism, significantly explain (p < .05)direct water scarcity risk ratings.
The Indirect Water Scarcity Risk multiple regression model results provided in table 3 indicated that the twelve risk perception factors explained 82.6% of the variance (adjusted R 2 = 0.826, R 2 = 0.917, F(12, 11) = 10.13,p < .001).Eight out of the twelve factors significantly explain (p < .05)indirect water scarcity risk ratings and include Discipline of Education (Natural Sciences), Discipline of Education (Arts), Discipline of Education (Others), Values_Egoistic, Values_Economic Benefits, Values_Sustaino-centric, Trust in Private Sector, and Attitude_Bipohysical Controllability.The Social Water Risk multiple regression model results in table 4 indicated that the ten risk perception factors explained 60% of the variance (adjusted R 2 = 0.60, R 2 = 0.774, F(10, 13) = 4.45, p = .007).Six out of the ten factors significantly explain (p < .05)social water risk rating and include Trust in Government, Trust in Private Sector, Discipline of Education (Arts), Attitude_Bipohysical Controllability, Knowledge_Self, and Knowledge_Experts.
Using multiple linear regressions, we also examined Trust in Private Sector (to manage water risks) and Overall Confidence in freshwater abundance in Ontario (as Dependent Variables).The Trust in Private Sector multiple regression model (table 5) indicated that the risk perception factors explained 60.6% of the variance (adjusted R 2 = 0.606, R 2 = 0.778, F(10, 13) = 4.54, p = .006).Six out of the ten factors significantly explained (p < .05) the degree of trust in the private sector, including Values_Egocentric, Gender, Attitude_Social Controllability, Impact_Financial, Knowledge_Self, and Knowledge_Experts.The results for Confidence (table 6) indicated risk perception factors explained 67.7% of the variance (adjusted R 2 = 0.667, R 2 = 0.846, F(12, 11) = 5.02, p = .006).Nine out of twelve factors significantly explained (p < .05)confidence, including Values_Technological Optimism, Drivers of Water Issues_Micro, Cultural Importance, Gender (Woman), Discipline of Education, i.e., Arts, Natural Sciences and Others, Location Conflict Rating, and Exposure.
To validate the assumed normality, linearity, and homoscedasticity for the linear regression model, the probability plots of regression residuals and scattered plots of predicted values with standardized residual values were visually assessed (Slimak andDietz 2006, Pearson 2010).Additionally, post hoc Shapiro Wilk Tests of the residuals (i.e., p > .05if residuals are normally distributed) and Breusch Pagan Tests for residuals were performed (i.e., p > .05for homoscedasticity) (Breusch andPagan 1979, Norman 2010).To overcome overfitting and model adequacy concerns, we calculated the adjusted R 2 values as well as the PRESS statistics, which should be lower than the model's total sum of squares (SSTO) (Allen 1971).All regression models upheld statistical assumptions, all VIF values were lower than three, and no multi-collinearity issues appeared (Slimak andDietz 2006, Mumbi andWatanabe 2020).
To reveal statistically significant mean differences based on gender and sector, we conducted the Student's ttest and an ANOVA, respectively.For the t-test on gender, we tested differences in three water risk perception factors (as dependent variables) Nature of Risk, Knowledge (self-assessed, general issues, and expert), Drivers of water issues (macro, micro), Water Attitudes (Scope, biophysical controllability, social controllability, concern, confidence) Impacts (reputational, financial, sustainability), Values (sustaino-centric, economic benefits, egocentric, technological optimism), Cultural Importance, Trust (private sector, government, civil society), Exposure, and Degree of Water Risk Integration.Results reveal that only two factors had statistically significant mean differences, i.e., We employed a one-way ANOVA to test sector-based differences for some variables of interest like the Water risk ratings, Impacts (reputational, financial, sustainability), Drivers of water issues (macro, micro), Trust (private sector, government, civil society), and Degree of Water Risk Integration.Results presented in table 7 demonstrate a statistically significant influence of the sector on the Macro drivers of water issues, F(7, 16) = 3.97, p = .011,Reputational Impact of water issues, F(7, 16) = 2.99, p = .033,Sustainability Impact of water issues, F(7, 17) = 5.85, p = .001,Degree of Water Risk Integration in organizational decision-making, F(7, 17) = 3.39, p = .019.Effect sizes (eta squared) in all cases were > 0.14.We performed the Shapiro-Wilk Test for normality, and sector differences were further explored in the interviews.

Stage 2 interview: guide preparation, thematic analysis, and results
After statistically analyzing the survey data, findings were categorized as new, expected, or unexpected based on the literature.New and unexpected findings were used to develop the follow-up one-on-one interview guide, enabling further exploration and interpretation of the survey's findings.

Interview guide design and procedure
The interview guide (Appendix E) included two questions aligned with RQ 1, which was intended to explore the conceptualization of different water risks as well as gain explanatory insights on the mean risk ratings found in the survey.Aligned with RQ 2, five questions were based on explaining interesting/unexpected risk perception factors or relationships found in the survey including Value of technological optimism and DV 1, Trust in private sector/industry and DV 2 and Knowledge_Self and DV 3.Moreover, we also discussed questions on exposure to water issues, proximity bias, role of trust, sector-specific insights, and gender-based differences.
All interviews were administered online on MS Teams following a standard interview protocol for consistency (Creswell and Creswell 2018).One research team member conducted all interviews to eliminate inter-interviewee bias.The questions were asked sequentially, but given the semi-structured nature, there was flexibility for change, e.g., additional prompt questions for elaboration if a new concept emerged.All interviews were audio recorded with 817 min of data for 22 participants.On average, interview data of 37 min per participant were collected, highlighting sufficient details and in-depth engagement (Money 2014b).All interview audio files were transcribed into separate documents with timestamps, anonymized for confidentiality, and reviewed for errors.

Qualitative coding and thematic analysis
We employed QSR NVivo 12 to code and analyze the interview data.We followed the coding and analysis procedures by Braun and Clarke (2006) for a priori theory-driven coding apt for explanatory sequential mixed methods.Starting with Pass 1 Open Coding, 22 transcripts were reviewed line-by-line and assigned codes or labels to capture the confirmatory or emerging ideas.The codes were named and defined based on the underlying concept (predetermined or expected codes) aligned with our theoretical framework and Stage 1 findings.Data/extracts could be coded under multiple codes to address the concept from various theoretical angles as necessary (codes of conceptual interest) (Braun andClarke 2006, Creswell andCreswell 2018).For Pass 2 Coding, Pass 1 codes were refined and condensed further, and some open-ended codes were retained for further conceptual organization.
For thematic analysis (Axial), Pass 2 codes were reviewed to identify conceptual patterns, and related codes (and their data extracts) were collated into higher-level themes.The themes were refined, defined, and named based on their underlying core concepts (Braun and Clarke 2006).For the final analysis (Selective), the themes were connected to the theoretical framework and hypotheses (factors of risk perception) and further collated into relevant sub-categories and main categories (Mooney et al 2020).The broader level of analysis was based on the main concepts drawn from the data, relationships between concepts, and aligned with (answering) the research questions (Mooney et al 2020).179 Pass 2 Codes (post phase 2 refining) were collated into 69 Themes under 20 Sub-categories that were further collated under 7 Main Categories.One research team member coded the data, and other team members independently cross-checked the codes and themes (Creswell and Creswell 2018).The analysis of water risk management strategies and next steps will be a part of future work.
Appendix C illustrates the thematic analysis for a single sub-category (Types of Water Risks) under the Nature of Risk category, along with the codes, the number of files (participants) that were coded under the theme, and an example of data extraction from the transcript.

Stage 2 qualitative results
Key findings from the thematic analysis are reported below, and the themes are italicized.A conceptual map was also developed to visually depict the resulting categories, sub-categories, themes, and sub-themes (Appendix D).

Nature of water risk
The introduction questions were exploratory and provided the participants' baseline conceptualization of water risks.Analyses of the responses revealed themes around the different types of water risks, the degree of water risk integration in organizational decision-making, drivers or barriers of water risk integration, prioritization, and management.The first theme of the discussion focused on different types of water issues and risks, where there was a conceptual distinction not only between biophysical aspects and social aspects of water risks but also among biophysical aspects of water quantity and quality.Moreover, temporal importance for water quantity as well as quality (via legacy contamination), location, i.e., spatial variability of water risks across sub-watersheds, and interconnection between quality and quantity emerged as important sub-themes.Nonetheless, the systemsbased interconnection between water quality and availability, i.e., quality-driven scarcity, was discussed only by 32% of the participants.
Social water risks also emerged to be multi-dimensional, where themes of conflict risks, regulatory risks, and the perceived impact of water-intensive sectors (e.g., water bottlers, agriculture, food processing, etc,) were discussed.Within social conflict risks, sub-themes like barriers to access to water for social, economic, environmental, or drinking water uses, competition among water users, nexus of water and health, concerns around legacy water-related inequity issues in First Nations communities, as well as negative media attention and public concern emerged.Confirming the factor analyses, participants articulated water risks as a multidimensional concept with biophysical and social facets further reflecting the nuances of public perception, temporal uncertainty, spatial variability, and sector-specific impacts that have implications for water risk evaluation.Among different risks, water quality is discussed more overtly by participants (91%).Source type risk has been conceptualized as the sensitivity of groundwater to contamination, aquifer productivity, competing demand, and access to the Great Lakes.
The discussion on the extent or degree of integration of different water risks revealed two key themes.50% of the participants mentioned significant consideration of water risks and their impacts at the individual organizational level.However, 50% of the participants (mainly from the financial sector) mentioned that water risks were integrated but to a limited extent at the organizational level.Participants revealed six potential water risk integration, prioritization, and management drivers.Firstly, organizational-level drivers focused on alleviating risk to business operations, the impetus of investors, and the individual organization's values and commitment to water-related sustainability targets, goals, assessments, and reporting.A connected yet distinct theme was of reputational drivers, where water risk integration is driven to avoid negative reputational impacts due to media, public perception, or water user conflicts, as well as to meet stakeholder expectations to maintain the social license to operate.From an institutional perspective, government regulations and compliance requirements were also identified as drivers along with economic and financial drivers like increasing costs and financial impacts of water issues.Finally, biophysical drivers, such as local water scarcity, contamination, or extreme climate events, followed by drivers related to individual concern and motivation were also identified.
Three key themes emerged when barriers to water risk integration were discussed stemming from the limited extent of water risk integration.Firstly, a lack of systems understanding about water was identified, where participants noted a lack of awareness about the interconnected nexus of water risks, cumulative impacts, efficiency of water use, and articulation of the business case of sustainable water use.Secondly, siloed business-asusual approaches and inertia was identified, where participants highlighted the inertia in organizations that tend to focus on the financial bottom line and metrics.Under the inertia theme, institutional issues like lack of transparent, integrated, and attuned or systems-based regulations that drive action and support sustainabilitybased transitions were highlighted.Moreover, the perception of water abundance in Ontario is discussed as a driver of the status quo.Finally, participants noted that water tends to be a lower priority/ focus than carbon and climate change.

Different water risk ratings and their reasons
Aligned with the survey, 73% of the participants discussed direct scarcity or water quantity risks being high in Ontario.Nonetheless, 77% of the participants reflected on water risks other than quantity, revealing four themes.Firstly, there is a perception of water abundance in Ontario, where 41% of the participants mentioned water quantity as relatively secure.Nonetheless, the nuances of variability due to location (proximity to the Great Lakes) or sector-specific differences in water use, were articulated in tandem.Secondly, participants also noted water quality risks, including the impact of wastewater contamination, to be equally high and prevalent across Ontario.Thirdly, 32% of the participants considered social water risks to be more relevant and challenging for the context of Ontario.Lastly, participants also noted a high risk of regulatory uncertainty due to evolving waterrelated regulations as well as increasing timelines, scrutiny, and level of detail required for water-related permit approvals.
Discussion on the reasons for the assigned water risk ratings revealed three key themes.Firstly, cognitive knowledge and awareness-based factors were identified by 86% of the participants, where awareness, knowledge of financial impacts of water issues, perceived impact of certain sectors on water resources, water availability of specific locations, the role of timing, demand as well as the influence of information on issues outside of Ontario (e.g., in the United States), were identified as reasons for higher risk perception and ratings.Moreover, the gap in knowledge and understanding about specific water issues, e.g., emerging contaminants and direct experience of an adverse event (instead of likelihood) were also identified as cognitive factors leading to high-risk perception.Secondly, internal attitude factors related to water (controllability, importance, or concern) were identified, where issues of competing water user groups and interests, as well as increasing unpredictability and controllability of water issues beyond one's sphere of influence/control are noted to shape the perception of water risks.Moreover, given the criticality of water, there is higher sensitivity regarding water availability and, hence, higher risk perception.Thirdly, the participants also attributed the reasons for risk ratings to Trust-based factors, where media coverage on water issues inside or outside Ontario can spur concern, distrust, and impact risk perception.Moreover, past reactive regulations related to water may create distrust, concern, and higher perceived risk that may be different from actual risk.

Explanation of risk perception factors and risk ratings based on survey findings
Firstly, the survey's findings (table 2) revealed that the Direct Scarcity Water Risk rating increases with technological optimism.We wanted to investigate this possible value-behaviour disconnect since optimism in technology did not alleviate the perception of direct scarcity (quantity) risk.82% of the participants mentioned the cognitive aspects of limits to technology related to water quantity risks.Under this broad theme, participants identified limitations of relying only on technology due to complexities of timing, demand, biophysical barriers like resource availability, economic barriers, and risks of novel technology that may not be tried and tested.Participants noted that gaps in knowledge and understanding about water quantity issues persist, i.e., scientific uncertainty due to which technology cannot fully mitigate the risk, leading to higher risk perception/ ratings.Moreover, the availability of technology is one aspect, but the lack of incentives and regulatory signals to adopt technology is another driver of risk perception.Participants also noted that there may be higher optimism and the possibility of technology to address water quality issues.Still, in the case of water scarcity or reduced flows, technology may have more barriers.Moreover, increased awareness/exposure to competing information, data, and technology about an issue may also lead to an increased risk perception.
Secondly, the survey's findings (table 3) revealed that the Indirect Water Scarcity Risk rating increases with the level of trust of an individual in the private sector (to self-regulate and manage water risks).Since our participants were from the private corporate or financial sector, we wanted to understand why a higher trust in the private sector does not alleviate perception and rating for indirect scarcity risk, i.e., water quality and source type sensitivity.68% of the participants noted a distinction between trust in one's organization, suppliers, or sphere of influence and trust in the broader industry beyond one's sphere of control or influence.Therefore, thematically, the results suggest that trust may exist in one's own organization and sphere of influence, but trust may not be generalized to the entire industry.So, while there is trust to an extent in one's organization to selfregulate, this trust is not extended to all industries without adequate checks and balances.Secondly, 73% of the participants discussed an affective disconnect where the presence of trust does not alleviate the issue.While there may be trust and action by the private sector, the indirect water scarcity or quality issues remain a higher priority.Various sub-themes emerged from this aspect, where the role and reliance on government regulations and laws to drive action and compliance in the private sector was highlighted to alleviate risk.Moreover, the self-serving and profit-centric interests of the private sector and the implications of other organizations or sectors that may not be sufficiently self-regulating, may still drive risk.Finally, the complex interaction of information and trust is articulated, where excessive information and negative media coverage tend to increase perceived risk and lead to tacit distrust.
Thirdly, from the survey (table 4), we found that Social Water Risk rating increases as one's own knowledge and professional experience related to water increases.Three key themes emerged from the discussion that explain an increased focus on the social aspects with experience.82% of the participants mentioned the role of external drivers of social aspects like stakeholder awareness, expectations, reputation, regulations, sector-specific impacts, and media, which become more evident with experience.Specific sectors are more prone to social water risks due to the nature of water taking, controversies, and media attention on these sectors compared to biophysical risks.Moreover, certain social aspects like stakeholder engagement and consultation are part of regulations, and broadly, there is more awareness regarding social aspects like the social license to operate.68% of the participants also highlight that at an individual level, the connection, awareness, and notions of controllability about social aspects of water risk increase with one's experiential knowledge and awareness.Finally, 41% of the participants also noted that tangible biophysical and economic aspects tied with social aspects might drive the emphasis on social water risks, where biophysical and social risks are increasingly understood to be connected and equally important.

Affective, spatial, and socio-cultural demographic risk perception factors
Participants also provided insights on other theoretically relevant factors like the role of previous exposure, gender, proximity bias, and trust in various institutions and their influence on water risk evaluation.Firstly, the discussion on previous exposure to water issues and its impact on overall confidence, concern, and risk perception revealed four key themes.77% of the participants highlighted that direct exposure to water issues increases water awareness, value, and importance.Thus, any exposure to or experience with local water issues like scarcity or quality issues, a first-hand realization of climate change impacts, as well as increasing competition and demand by users could raise awareness and hence relevance and risk perception of water.Participants also noted that professional experience and disciplinary knowledge related to water increase risk perception.The disciplinary understanding gained in educational training and the awareness and experiences gained in a professional setting can make an individual more attuned to water issues, increasing risk perception and concern.Another theme addressed the exposure to non-local water issues that increase the value and confidence of water quality and quantity in Ontario.In this case, participants articulated that experiences or information about water issues in other regions or countries may also increase the value and relative confidence in water security in Ontario.Finally, the theme of impact of non-water issues and experiences highlights that some non-water issues, like inflation, food and fuel prices, etc, may garner higher attention than water.Moreover, negative experiences like regulatory stringency or enforcement may generate a defensive reaction.
Secondly, upon exploring gender and risk perception with a subset of participants who identified as women, two broad themes emerged.The first theme highlighted that women tend to have higher empathy, awareness, and vulnerability to impacts, hence are more concerned about access to the environment and water.Due to social norms historically, women tend to be more empathetic, aware, and considered as nurturers of families and, broadly, the environment.Moreover, women are increasingly in more front-and-center decision-making roles related to sustainability and water, using collaborative approaches instead of relying on technology alone.Marginalized groups, including women, tend to be more vulnerable to adverse impacts of environmental issues, and this may translate to lesser confidence in freshwater availability, higher concern, and perceived risk.
Participants also mentioned that water risk perception may also be influenced by other considerations like the intergenerational concern of people with families or people with professional roles related to the environment.
Thirdly, the discussion on proximity bias for water issues and risk perception, i.e., potential discounting of water risks in one's sub-watershed, revealed three themes.The first theme focused on the perception of more control over local water issues, where more engagement with local water issues and proximity to the Great Lakes may lead to the perception of higher control over local issues and reduce risk perception.Moreover, from a cognitive perspective, participants discussed how there might be a general lack of knowledge about local water issues and higher knowledge of the impact of non-local water issues that may increase concern/ perceived risk outside one's region.The media and public attention on the broader Great Lakes issues like nutrients, water diversion, etc, is more than local water issues.Hence the concern for local water issues may be lowered.
Finally, the survey revealed (section 4.1) that there was, on average, higher trust in civil society organizations and government to assess and manage water risks.In the discussion, 82% of the participants attributed lower trust in the private sector to its profit-centric and self-serving interests.77% of the participants perceived government and civil society organizations public-centric and neutral institutions, working for public interests without profit motives, leading to higher trust.Moreover, the experiential theme of past demonstration, systemic tendencies, and history of water management also emerged.Participants noted that government and civil society organizations may have demonstrated expertise, collaboration, and transparency, along with considering cumulative impacts leading to higher trust.Nonetheless, trust may taper down due to past negative experiences as well as knowledge/ understanding of the profit-centric motives and self-interests of the private sector.Finally, a lack of transparency in the private sector may lead to higher trust in non-private actors.Participants also highlighted the role of media, where the nature of messaging may create distrust in the private sector.

Sector-specific insights on water risk ratings
Contrary to the hypothesis, H14 Sector , there were no significant sector-based differences for the three water risk ratings from the statistical analysis.However, from the interviews, discussion regarding sector-specific differences in water risk priorities revealed nuances of this relationship.Firstly, 41% of the participants noted that there tend to be some sector-based differences in the ranking or prioritization of different water risks due to variability in water use across sectors in terms of water quantity and quality.Nonetheless, they elaborated high heterogeneity of water use even within sub-sectors that other sectors or industries may not completely understand.32% of the participants emphasized that the location of water use is more important than the sector, where the location may be more relevant in the water risk ratings than the sector.Moreover, participants also noted that water risk ratings are contingent on the sector and the location, further highlighting the complexity of sector-based influences on risk ratings.
Discussing different impacts as drivers of how various water risks are rated, prioritized, or managed across various sectors revealed three themes.Firstly, participants noted the influence and relevance of the financial impact of risks as the most important (compared to sustainability or reputational impacts) in driving differences in water risk priorities across different sectors.Thus, the extent of the negative financial impact of water issues may be the most critical factor leading to different water risk priorities/ratings across sectors.Secondly, the participants noted that regulations and compliance may drive a sector's water risk priorities, especially if the sector (e.g., chemical manufacturing, food and beverage, mining, etc.) is more regulated than others.Finally, reputational, social, and sustainability impacts may drive sector-based differences, where the extent of negative impacts on sustainability and reputation rather than just financial implications may lead to differences in prioritization of risks.

Discussion of results
Our study's findings provide several interesting insights that contribute to the literature on water risk perception, assessment, and management in the CFS.Our study's novelty and contribution to knowledge is the empirical examination and explanation of the complex construct of water risk perception and its relationship with risk evaluation by applying the interdisciplinary Risk Theory to the water domain in the CFS, which has not yet been addressed in the literature.The application of explanatory sequential mixed methods is a major strength of our study that led to enhanced scientific rigor, qualitative depth, and nuanced theoretical understanding of complex constructs, such as water risk perception and evaluation.
The following sub-sections discuss the results with regard to our study's main objectives: firstly, to empirically examine the factors underpinning water risk perception and priorities of practitioners and decisionmakers in the CFS in Ontario.Secondly, to explain the relationship of water risk perception with risk evaluation.We discuss the results from both stages based on the study's hypotheses and the current literature.Finally, we discuss the study's contribution to knowledge and practice along with the limitations and recommendations for future work.

Factors underpinning water risk perception of the CFS
Addressing RQ 1, our study found technical characteristics of risk as well as more individual-centric cognitive, affective, socio-cultural, and spatial factors that underpin water risk perception of corporate and financial practitioners, analysts, and decision-makers and influence water risk evaluation.From the perspective of the risk object, i.e., water risks, the factor analysis of the ratings of different water risks (section 4.2) and interviews revealed a conceptual distinction between biophysical and social water risks.This distinction was interesting because it confirmed the multidimensionality of water risks as a concept and in decision-making with both biophysical and social dimensions.Thus, supporting H1 Nature of Risk , heterogeneity among and within different biophysical and social water risks was confirmed.However, in contrast to Money (2014a) who generated their results from corporate reports, our study used a mixed-methods approach including a survey and follow-up interviews of practitioners and decision-makers that are able to assess individual experiences.Moreover, in contrast to Mumbi and Watanabe (2020), our study focused on Ontario-a seemingly water abundant region, where contemporary water issues emerge locally.Hence, our study broadened the literature through the examination of water risk perception of individual practitioners and decision-makers in different water-using sectors and by using a granular lens on a seemingly water-abundant region as the field of study.
Aligned with the study's theoretical framework, we find that the selection, interpretation, and prioritization of various water risks or framing of water risks is contingent on technical aspects of the risk object but also on the individual's expertise, knowledge, or exposure to issues, internal values and attitudes related to water as well as trust.Our results suggest that the latter is valid for practitioners (Dobbie andBrown 2014, Mumbi andWatanabe 2020), analysts, and decision makers in the CFS in Ontario (Klinke andRenn 2012,2019).Interestingly, from the interviews, the reasons for the assigned water risk ratings were attributed to cognitive factors like awareness, direct experience with the issue, perceived or known impact of other sectors on water resources, and knowledge of the extent of impacts.Moreover, affective aspects like values and attitudes related to water, i.e., controllability, intrinsic connection to water, perceived criticality, nuances of location, and trust related to media, stakeholders, and perceived sufficiency of regulations, were also articulated.Hence, the results address our first objective by suggesting that a nexus of cognitive and affective factors underpin water risk perception and priorities of practitioners in the CFS, where awareness or knowledge (tacit or acquired), attitudes, and trust (in different institutions to manage water risks) shaped water risk perception.
While sector-based differences in the three water risk ratings were not statistically significant, interviews revealed additional insights.Heterogeneity among sub-sectors, nuances of location, and risk impacts may be difficult to capture using quantitative methods alone.Nonetheless, in line with Dobbie and Brown (2014) and Klinke and Renn (2019), the extent of integration of water risks in decision-making were found statistically to vary across sectors (table 7).From the interviews we found that the extent of water risk integration in decisionmaking was contingent on individual organizations rather than sector-wide generalization.This emphasizes the need for surveys and interviews at the individual level to assess water risk perception.From the discussion on the drivers for integration, we found a dominance of institutional and stakeholder-related drivers like reputation, organizational risks (operational institutional drivers related to stakeholder risks, investors, etc.), regulations, and economic drivers rather than pure biophysical drivers of local water security in line with the water governance findings reported by Alvarado-Revilla and de Loë (2022).
Additionally, the financial sector tends to be more critical of the limited extent of water risk integration across all sectors.Barriers to water risk integration were related to cognitive gaps in systems understanding of water but also the inertia of business-as-usual approaches.This inertia was attributed to a lack of clarity regarding the business case or economics of sustainable water management, a lack of integrated and attuned regulations, and path dependency on financial metrics and the bottom line.Also, evidence of 'carbon tunnelvision' emerged, where participants articulated the myopic and competing focus on carbon emissions, energy, and climate that overlooks the connection between water, carbon, and climate (Konietzko 2022).Based on these results, we suggest that transdisciplinary collaboration, training, and knowledge sharing is needed to foster proactive systems-based approaches for assessing and managing different water risks.
Other interesting risk perception factors in our factor analyses include participants' Concern about water.Supporting H4 Concern , results (section 4.2) revealed that factors like the urgency of water issues, perceived inequity of issues, and non-substitutability of risks by economic benefits contribute to overall concern related to water issues.Hence, it seems that the participants preferred a strong sustainability approach to water (Dietz and Neumayer 2007).The construct of Confidence (in abundance and sufficiency of water resources in Ontario) was also interesting.Supporting H13 Cultural Importance , the cultural importance of water was a significant explanatory variable negatively related to confidence (table 6), where higher cultural importance of water tends to decrease overall confidence.This was expected as one's shared cultural beliefs and attitudes connect to higher sustainocentric values emphasizing protective behavior and preparedness related to water conservation intra and intergenerationally instead of confidence, aligned with Dupont et al (2014) and Siegrist (2019).However, contrary to H5 Exposure , exposure to water issues increased confidence in freshwater abundance.Theoretically, an increased risk perception, concern, protective behaviour, and decreased confidence was expected (Thistlethwaite et al 2018, Siegrist 2019).Nonetheless, the interviews revealed that previous exposure to water issues can evoke complex reactions that are contingent on whether the experience was direct or indirect through information about water issues that affected others.Moreover, non-local water issues may increase risk perception and concern but also tend to increase the value and confidence in the relative availability of water resources in Ontario.This result explains the perception of Ontario's relative abundance of water.
Gender was an interesting demographic variable, in which statistically significant mean differences were found using a t-test for technological optimism and overall confidence.From the interviews, women were more concerned about risks to human health, environment, intergenerational well-being, and safety due to the social gender norms of being entrusted with nurturing families.While studies by Slovic (1999) and Dupont et al (2014) found similar results for perception of other risk domains and Mumbi and Watanabe (2020) report no significant effect, our study confirms the role of gender in the water domain and the specific sample of the CFS in Ontario.Marginalized groups, including women, tend to be more vulnerable to the adverse impacts of risks.Thus, similar to risk domains (Slovic 1999), women tend to be more sensitized to water risks, explaining the lower confidence in freshwater abundance, higher concern about water issues and their intergenerational impacts, and lower optimism in technology as the sole solution.

Relationship between water risk perception and risk evaluation
Addressing RQ 2, we developed multiple linear regression models and found statistically significant relationships between risk perception factors and water risk evaluation.Water risk perception of analysts, practitioners, and decision-makers in the CFS is highly nuanced, and the underlying explanatory factors/models vary with the type of water risk (Dobbie and Brown 2014).Interestingly, factors like knowledge, values, attitudes, and trust were important to different extents for the three types of water risks.As reported in tables 2, 3, and 4, the attitudinal factor of controllability, aligned with the Theory of Planned Behaviour and connected to familiarity and dread dimension, was a common significant explanatory variable in all three water risks (Weber et al 2002, Dobbie andBrown 2014).Thus, supporting H3 Attitude Controllability , higher perceived controllability of biophysical water issues was associated with a lower rating of water risks.Hence, the study clarified the role of the controllability factor in the Theory of Planned Behaviour specifically applied to water risk evaluation.
For the direct scarcity risk rating (table 2), supporting H6 Knowledge , Knowledge, specifically in experts about water issues, impacts, risks, and likelihood, was positively associated with direct scarcity risk rating.Hence, the study confirmed the results for water risks in line with the literature on the impact of knowledge on the perception of broader environmental risks (McDaniels et al 1997, Weber 2001, Slimak and Dietz 2006, Siegrist and Árvai 2020).However, one's own knowledge was not a statistically significant factor but two out of the four values were relevant, where supporting H8 Egocentric Value , a negative association between egocentric values and the risk rating was found, where the higher egocentric values of an individual, the lower the risk rating (Bouman et al 2018).Contrary to H10 Technological Optimism Value , higher technological optimism did not alleviate risk but increased the risk ratings.While technological optimism as a significant explanatory variable aligns with the Value-Belief-Norm Theory and Theory of Reasoned Action (Weber et al 2002, Slimak andDietz 2006), the unexpected direction alludes to a possible disconnect between one's values and behaviour (captured in risk ratings).
From the interviews, we found a tacit understanding of the limits of technology, i.e., biophysical limitations of resources as input and lack of economic and regulatory signals for adoption.The remaining factors supported our hypotheses, i.e., H2 Attitude Scope , the higher the perceived extent or scope, i.e., the area and people impacted by the specific water issue, the higher the direct scarcity risk rating.Moreover, supporting H12 Gender , women tended to rate direct scarcity risks higher than men, alluding to gender-based differences in the rating of specific water risks.Hence, our study confirmed that risk perception in the domain of direct water scarcity risk is shaped by individual experiences and subjective factors (Sjöberg 2002, Slimak andDietz 2006).
For the indirect water scarcity risk rating (table 3), based on the survey's findings, in contrast to Siegrist and Árvai (2020), we found no significant influence of one's own or expert knowledge/experience on the risk rating, alluding to the dominance of affective factors like values and trust in the CFS practitioners.Supporting H7 Sustainocentric Value , we found that higher biospheric and altruistic values tend to increase risk ratings as well as H9 Economic Benefits Value , where economic benefits centric values were positively associated with indirect scarcity risk, unearthing the connection between financial materiality and indirect scarcity risks.H8 Egocentric Value , was also supported where the higher egocentric values, the lower the indirect scarcity risk ratings.Interestingly, as compared to direct scarcity, a more diverse set of values explains indirect scarcity risk ratings, but technological optimism was not an explanatory factor.Nonetheless, the discipline of education was a significant explanatory factor, where participants with an educational background in Arts (including Economics, Business, Finance, etc,) tend to rate indirect scarcity risks higher than participants with a background in Environment, Natural Sciences, and Other disciplines.Thus, alluding to the role of educational discipline behind differences in risk perception of water quality and source type risks (Dobbie and Brown 2014).From the interviews, water quality risk is more important from a financial materiality (impact) perspective, especially in the financial sector.Overall, these results suggest a strong influence of affective and individual factors on water risk evaluation.Hence, the risk perception might be highly variable and different from biophysical parameters of water scarcity.
Finally, trust in the private sector/industry was a significant explanatory factor that tended to increase the indirect scarcity risk rating.Given that the sample is predominantly from the private sector, based on the literature, we expected value alignment and risk alleviation based on trust (Siegrist 2019, Mumbi andWatanabe 2020).Nonetheless, from the interviews, trust emerged as a nuanced construct operating in different spheres.Higher trust is placed within one's immediate sphere of influence or interaction, i.e., one's own organization, suppliers, etc, but not generalized across the sector or industry resulting in higher perceived risk.Thus, trust is limited to one's immediate sphere and it does not alleviate risk due to the perceived impact of all sectors.Moreover, trust was an explanatory factor in indirect scarcity risk but not in direct scarcity risk, where knowledge was important, confirming reliance on either knowledge or trust for water risk evaluation (Siegrist 2019).
For the social water risk rating, which was expected to be more contingent on affective values, none of the four values were significant explanatory factors (table 4).On the other hand, supporting H6 Knowledge , i.e., one's self-assessed knowledge, professional experience, and knowledge of experts increases the social water risk ratings.This finding is in line with studies on other types of risks that found a correlation between self-assessed knowledge and risk perception (McDaniels et al 1997, Weber 2001, Slimak and Dietz 2006).Thus, as the awareness and professional experience of the practitioner increases, there is a higher emphasis on social water risks.Moreover, supporting H11 Trust , higher trust in the government to manage water risk by regulations, tends to decrease social water risk ratings.Thus, higher reliance on the government to manage social water risks (competence-based trust) is evident (Siegrist 2019).Interestingly, knowledge and trust explain the evaluation of social water risks rather than intrinsic affective values.In the interviews, participants explained that cognitive and external factors like stakeholder awareness, expectations, nuances of power, reputation, regulations, sectorspecific impacts, and the role of media become more evident with professional experience.Thus, the study reveals a novel insight, where compared to biophysical risks, social water risks were more experiential and the awareness of complexities of social water risks increases risk perception and evaluation.Furthermore, these findings on social water risk perception in the CFS confirm and add to the findings by Alvarado-Revilla and de Loë (2022), who studied external factors influencing water quantity governance in Ontario.
Participants also provided insights on the survey's findings on proximity bias and trust and their interaction with concern, confidence, water risk perception, and evaluation.Supporting H15 Location , evidence for a proximity bias was found, where, on average, overtly stated concern for water issues in Ontario was higher than one's own sub-watershed (Krewski et al 2008).The interviews explained that a higher perceived locus of control and engagement in one's location (sub-watershed) tends to reduce concern and perceived risk that may be disconnected from actual biophysical water risks.Interestingly, discounting of water risks was also related to cognitive aspects of excessive information about non-local issues or lack of information about local issues, where non-local issues may take cognitive precedence in risk ratings.Since there is high media coverage regarding water issues (Sandhu et al 2023), this result on proximity bias and higher water risk perception for non-local issues is explained.This finding also adds to the literature about the role of media in risk perception and evaluation that has explored other risk domains (Flynn et al 1998, Slovic 1999, Kasperson et al 2022) by including the water domain in the CFS.
For trust in different institutions, results revealed that, on average, higher trust is placed in government and civil society organizations.Among private sector (non-state) practitioners, even given their critique on the design and reactive nature of water regulations, public-centric institutions are still predominantly relied upon to provide unbiased oversight, checks and balances, and drive compliance and, thus alleviate water risks.Trust in the industry/ private sector to manage water risks, as discussed before, was limited to one's sphere of influence and not generalized to the entire industry primarily because of the awareness of the self-serving profit-based interests of the private sector more broadly.Thus, aligned with the regression results for trust (table 5), cognitive factors like knowledge, awareness of impacts, as well as the role of media and messaging, and lack of transparency reduce trust in the industry, even among practitioners, thus increasing risk perception.Our study's results for trust and water risk perception is in line with the review by Siegrist (2019) covering other risk domains, where our study adds to the literature by examining the under-explored water domain.The role of media was a consistent theme for driving water risk perception and proximity bias and emerged as an influential institution that influences cognitive factors and affect due to its impact on trust (Slovic 1999, Mooney et al 2020, Kasperson et al 2022).

Study's novel contributions to knowledge and practice
The study makes an original contribution to knowledge by systematically examining and explaining the water risk perception of CFS practitioners and decision-makers and the relationship with water risk evaluation using explanatory sequential mixed methods, addressing a major gap in the current literature (Siegrist 2019, Mumbi and Watanabe 2020, Renn et al 2020, Di Baldassarre et al 2021, Dudley et al 2022, Rangecroft et al 2021, 2022).Firstly, our study expanded and empirically tested the novel interdisciplinary framework of Risk Theory for the water domain with a first-of-a-kind sample of non-state CFS practitioners and decision-makers at the regional scale (Klinke andRenn 2019, Dudley et al 2022).The study's novel findings highlight the importance of integrating psychometric, cultural, and relational theories of risk to unearth contextual and multidimensional complexities of risk perception in the assessment and management of water risks.The study concludes that for practically sound and comprehensive water risk management, in addition to quantifying water risks, it is equally important to integrate diverse perspectives of influential actors like CFS who are undertaking water-related decision-making (Christ andBurritt 2018, Klinke andRenn 2019).The study's findings have key implications in the fields of risk analysis and governance, sustainable water management, and decision-making.For instance, the interviews revealed that water quality risk tends to be important in the financial sector from a financial materiality (impact) perspective.Moreover, water risk integration in more regulated sectors, e.g., chemical manufacturing, food and beverage, mining, etc, tends to be driven by regulations and compliance.Thus, these diverse sector-specific nuances build a case for an attuned water risk management framework rather than a onesize-fits-all framework based only on biophysical risk data.
Secondly, the study finds that the water risk perception of CFS experts, i.e., analysts, practitioners, and decision-makers, was highly nuanced and shaped by cognitive, affective, socio-cultural, spatial, and trust-based factors that influence the evaluation of different water risks (Dobbie andBrown 2014, Kasperson et al 2022).Therefore, by including variables like gender, educational discipline, location, etc, as part of our novel interdisciplinary theoretical framework, we empirically confirmed their impact and importance for water risk assessment.Otherwise, omitting these theoretically relevant variables could have skewed the findings.Moreover, water risk was identified as a systemic risk problem with all three characteristics of complexity (due to a multitude of explanatory mechanisms), uncertainty (due to competing knowledge claims, experience, or lack of knowledge), and ambiguity (due to multiple interpretations, socio-political and normative value differences) (Aven andRenn 2019, Renn et al 2020).Consequently, there is a need for a variety of deliberative and discursive risk management, communication, and governance strategies to develop collaborative transdisciplinary knowledge of water risks among all stakeholders (Klinke and Renn 2019).Finally, methodologically, a key strength and scientific novelty of our study is the successful demonstration of the application of explanatory sequential mixed methods to examine complex constructs like water risk perception and evaluation in a particular expert sample, i.e., the CFS (Siegrist andÁrvai 2020, Di Baldassarre et al 2021).The follow-up interviews validated the survey's findings and provided much-needed depth and understanding of the drivers of risk ratings, expanding statistical evidence further.Therefore, using mixed methods, our study achieved enhanced rigor, qualitative depth, and theoretical understanding of nuanced water risk perception factors.
Practically, the study is part of a broader interdisciplinary project that aims to develop a locally attuned water risk management framework for the province of Ontario, Canada.Different hydrological and social water risks were investigated for Ontario in a separate study (Sandhu et al 2023).Our current study engaged CFS practitioners to co-develop knowledge and understanding of water risk perception and evaluation.Water risk ratings analyzed in this study can be used to design contextually-attuned decision support tools for the CFS as well as training strategies for risk assessment and management.Policymakers may find the results on proximity bias, the role of trust and exposure, the need of contextually-attuned regulations, the influence of media, location, and financial impacts as strategic avenues for multi-stakeholder engagement to bridge the sciencepolicy-practice gap in sustainable water management.

Limitations of study and recommendations for future studies
While apt for our explanatory research objective, the cross-sectional survey design does not provide causal interpretations.Nonetheless, the follow-up interviews provided insights into potential causal mechanisms like cognitive (knowledge/awareness), affective (attitude and trust-based), spatial, and sociocultural demographic factors that drive the water risk ratings.The study's findings inform future longitudinal survey designs to establish causality.Moreover, the scales for water attitudes and values developed in the study can be tested further with the lay public and experts in other public or civil society sectors for cross-sample reliability.A large sample size tends to be desirable in risk perception studies.However, given the time-intensive nature of mixed methods, large sample sizes are not always possible, especially with specific expert populations.Nonetheless, even with a specific and small sample, statistical assumptions were upheld, and reliable conclusions were drawn that were further validated by the interviews.
With respect to generalizability, our in-depth case study approach captures locally relevant heterogeneous water risks and perspectives of the CFS scoped to Ontario (Hogeboom et al 2018).Nonetheless, for future studies, we recommend applying our novel theoretical framework and mixed methods procedures to different geographical contexts for comparative case analysis, appealing to academia, businesses, investors, and policymakers beyond Ontario (Creswell and Creswell 2018).Secondly, future studies can operationalize our theoretical framework's water risk management stage and investigate management strategies and sector-based preferences to develop a decision support tool.Thirdly, future research can validate our study's framework with other stakeholder groups, including the public, government, and civil society organizations.Finally, future studies can further explore proximity bias, quantifying location-based differences between actual and perceived water risk ratings, operationalizing water risk governance frameworks, and exploring adaptive and resilience approaches for water risk governance (Klinke and Renn 2019, Kasperson et al 2022, Koehler 2023).

Conclusion
Extant literature revealed a pertinent gap in the empirical assessment of water risk perception and its impact on water risk evaluation and decision-making in the CFS.Our study addresses this gap in knowledge by examining the relationship between water risk perception and evaluation using novel interdisciplinary Risk Theory and explanatory sequential mixed methods approaches consisting of a survey and follow-up interviews.Our study finds heterogeneity among different water risks as well as multi-dimensional cognitive, affective, and sociocultural personal factors of water risk perception, including knowledge, professional experience, perceived controllability, values, trust, location, and gender of CFS practitioners, analysts, and decision-makers that influence the evaluation of different water risks.Water risk perception, evaluation, and decision-making is a nuanced and subjective process influenced by the risk characteristics and the risk perceiver.Therefore, for comprehensive and sound water risk assessment, management, and decision-making, we find it is critical to systematically integrate water risk perception and priorities of relevant non-state actors like the CFS along with quantifying different water risks.Our study contributes and advances knowledge in sustainable water management and risk analysis by empirically applying Risk Theory to the domain of water risks in CFS and successfully demonstrating the novel use of mixed methods for examining complex theoretical constructs like water risk perception and evaluation.Our findings on water risk ratings, relevant water risk perception factors, and sector-specific priorities enable the integration of analytical risk data with practitioner perspectives to inform evidence and context-based decision support tools for water risk management.These outcomes are critical for sustainable water management research, policies, and operational practices to promote water security and resilience in Ontario and Canada.

Figure 1 .
Figure 1.Theoretical framework for water risk perception, risk evaluation, and management.

Figure 2 .
Figure 2. Explanatory sequential mixed methods research design.

Table 1 .
Summary demographic statistics of the participants.
a One additional participant who participated only in the interview stage.
a b Dummy variable for the level of education (Reference: University Degree or Certificate, Master or PhD level).c Dummy variable for gender (Reference: Man).d Shapiro Wilk Test p = .941,Breusch Pagan Test p = .985.

Table 3 .
Regression coefficients, t value, significance and VIF for indirect scarcity water risk rating d (DV2).
a b Dummy variable for gender (Reference: Man).c

Table 6 .
Regression coefficients, t value, significance and VIF for overall confidence (attitude_confidence) d .

Table 7 .
Significant sector based group means and standard deviation.