Perceived Geotechnical Risk, Trust in Technical Assistance, and Willingness to Pay for Mitigation: A Cross-Sectional Study Among Residents and Construction Professionals in Cuenca, Ecuador

Abstract

Landslide and slope-instability hazards increasingly threaten residential areas in rapidly urbanizing Andean cities, where informal construction is common and formal geotechnical assessment is limited. Protection Motivation Theory and the Theory of Planned Behavior have been widely applied to explain protective intention toward natural hazards, yet few studies have jointly examined perceived geotechnical risk, trust in technical assistance, and willingness to pay for mitigation within a single model. This study assessed these constructs, together with mitigation adoption intention, among 420 residents and construction-related professionals in Cuenca, Ecuador, using a cross-sectional, descriptive–correlational design. Data were collected through a structured online questionnaire with seven Likert-scale constructs and a contingent-valuation item, analyzed using non-parametric correlation and group-comparison tests, followed by confirmatory factor analysis and structural equation modeling. Perceived risk, trust, and mitigation intention were all high and positively associated, while economic constraint correlated positively, rather than negatively, with intention, an effect that a structural model showed to be fully mediated through intention instead of acting directly on willingness to pay, which points to a recognized structural barrier and not an individually suppressive one; attitudinal willingness to pay did not consistently predict the declared monetary amount, and civil engineering professionals reported higher willingness to pay than other groups. These findings indicate that respondents in this sample reported high risk awareness and trust in technical assistance, while perceived economic constraint was the dimension most consistently associated with lower stated readiness for geotechnical mitigation in this Andean urban context. These findings contribute empirical evidence on the perceptual, trust-related, and economic drivers of household engagement with geotechnical mitigation, informing sustainable, resilience-oriented approaches to reducing disaster risk in informally built urban settlements across the Andean region.

1. Introduction

Geotechnical hazards, including landslides, differential settlements, and slope instabilities, remain among the most persistent threats to residential infrastructure worldwide, particularly in regions undergoing rapid or unplanned urbanization. Community resilience to natural hazards is understood to depend not only on physical exposure but on place-specific antecedent conditions—social, economic, institutional, and built-environment characteristics—that shape how a community absorbs and recovers from a hazard event [1]. Within this place-based framework, housing conditions and land-occupation patterns are recognized as key determinants of vulnerability, frequently confining lower-income households to marginal, geotechnically unstable land where housing remains exposed to landslides, slope failures, and related ground-related hazards [1]. This global pattern is especially pronounced in mountainous regions, where steep topography, seismic activity, and intense rainfall converge to increase both the frequency and severity of slope failures. Consequently, understanding how populations living under such conditions perceive geotechnical risk, and whether that perception translates into protective action, has become a central concern for disaster risk reduction research. Nowhere is this concern more relevant than in the Andean region of South America, where geomorphological conditions and accelerated urban growth converge to produce some of the highest landslide susceptibility levels in the world.

The Andean region is recognized as one of the most landslide-susceptible areas in the world, a condition driven by steep terrain, active tectonics, and decades of accelerated, often informal, urban expansion [2]. Empirical studies conducted across Andean provinces have identified urbanization, poverty, and socioeconomic inequality as key factors that increase landslide vulnerability, particularly where settlements occupy slopes, ravines, and other geotechnically sensitive terrain [3]. In the specific case of Azuay Province, where the city of Cuenca is located, landslide susceptibility mapping has confirmed that a substantial share of the provincial territory presents conditions favorable to slope instability, reinforcing the relevance of geotechnical risk as a local, and not merely regional, concern [3]. Beyond the physical susceptibility of the terrain, land-policy research in Andean cities has emphasized that landslide risk reduction depends not only on engineering interventions but also on the socioeconomic and institutional dimensions that shape how residents build, occupy, and modify land over time [2]. This multidimensional framing suggests that risk reduction in Andean urban contexts cannot rely solely on hazard mapping or infrastructure investment, but must also account for the perceptions, decisions, and constraints of the households that inhabit at-risk terrain. In Ecuador specifically, these dynamics are compounded by a construction sector in which a substantial proportion of residential housing is built informally, often without the technical guidance required to manage geotechnical risk appropriately.

Informal, self-built housing constitutes a substantial share of the residential stock in Ecuador, frequently constructed without prior soil studies, structural design, or professional supervision. National case evidence indicates that, in areas of high seismic and geotechnical risk such as Santa Elena Province, approximately 70% of housing has informal origins and exhibits structural deficiencies, reflecting a systemic gap between hazard exposure and the technical assistance available to homeowners [4]. This pattern is consistent with broader evidence from self-built housing across Latin America, where construction decisions are frequently guided by empirical knowledge and cost considerations instead of formal geotechnical or structural assessment. Such practices are particularly consequential in cities like Cuenca, where residential expansion has historically occurred over terrain that includes fill material, steep slopes, and drainage-sensitive ravines, conditions directly associated with settlement, cracking, and slope failure in housing. This mechanism is well documented in the geotechnical engineering literature: lightweight residential structures founded on reactive or poorly characterized ground are particularly susceptible to damage from ground movement, and the resulting financial losses can be substantial when design does not adequately account for site-specific soil behavior [5]. Despite this documented physical vulnerability, comparatively little research has examined how residents and construction-related professionals in Ecuadorian cities actually perceive geotechnical risk, or whether they trust and are willing to engage with the technical assistance required to mitigate it. Comparable survey-based approaches have documented gaps between physical vulnerability and community risk perception in other developing-country urban contexts, such as Peshawar, Pakistan, reinforcing the relevance of collecting similar evidence for Andean cities exposed to geotechnical hazards [6]. Addressing this gap requires a behavioral, and not merely physical, framework capable of explaining why risk awareness does not automatically translate into preventive action.

Protection Motivation Theory (PMT) and the Theory of Planned Behavior (TPB)—within which attitudes toward a protective behavior can be decomposed into instrumental and experiential components that independently predict behavioral intention [7]—constitute two of the most widely applied frameworks for explaining how individuals evaluate threats and decide whether to adopt protective measures [8,9]. Under PMT, protective intention results from the interaction between threat appraisal, comprising perceived severity and vulnerability, and coping appraisal, comprising perceived response efficacy, self-efficacy, and response cost [8]. Applications of PMT to landslide and flood contexts have repeatedly shown that high risk perception does not necessarily produce high protective intention, particularly when households lack confidence in their own capacity to act or perceive the recommended measure as costly or difficult to implement [10,11]. This pattern is consistent with two independent meta-analytic reviews of the PMT literature, which converged in identifying coping appraisal variables, and self-efficacy in particular, as the strongest and most consistent predictors of protective intention and behavior across health and safety domains [12,13], while a parallel meta-analytic review of TPB found that perceived behavioral control explains variance in intention and behavior independently of attitude and subjective norms [14]. Recent studies published in MDPI journals have integrated PMT with TPB into a single explanatory model, showing that attitude, subjective norms, and perceived behavioral control operate alongside threat and coping appraisal to shape disaster-preparedness intention across flood-prone and multi-hazard urban contexts [15,16,17,18]. This pattern, at times described as a “fatalism trap,” has been documented even among populations that are both aware of landslide risk and directly exposed to it, suggesting that perceived response cost can neutralize an otherwise strong risk awareness [11]. Within this framework, trust in the technical actors responsible for delivering a protective response, such as engineers or geotechnical specialists, functions as an additional condition for translating risk perception into mitigation intention, yet it remains comparatively underexamined relative to threat appraisal itself. Because response cost is, in practice, most often experienced as a monetary constraint, understanding the willingness of at-risk populations to pay for technical geotechnical assistance becomes a necessary extension of this theoretical discussion.

Willingness to pay (WTP), typically estimated through contingent valuation methods, provides a direct behavioral indicator of how much economic value individuals assign to reducing a given risk, and has been widely applied to health, environmental, and agricultural hazards [19]. Evidence from contingent valuation and survey-based research indicates that WTP for risk reduction is shaped not only by household income but also by risk perception and, notably, by the level of trust that respondents place in the actors or institutions responsible for delivering the risk-reducing service, a pattern documented both in agricultural risk contexts and, more directly, in citizens’ willingness to pay for disaster management activities [19,20]. This trust-dependent pattern has been documented primarily in agricultural, public-health, and general disaster-management contexts, while its extension to geotechnical risk mitigation, a domain in which the “service” being valued is a technical inspection or engineering recommendation instead of a policy or product, remains largely unexplored. In the specific context of Andean cities such as Cuenca, no study to date has jointly examined perceived geotechnical risk, trust in technical assistance, and willingness to pay for minimal geotechnical mitigation measures within a single behavioral model. This gap is particularly relevant given that response cost, and specifically its economic dimension, has been identified as one of the central barriers preventing the translation of risk perception into protective action in comparable hazard contexts [10,11]. Addressing this gap requires empirical evidence capable of simultaneously characterizing residents’ risk perception, their trust in technical assistance, and their disposition to pay for it, within a single, geographically specific case study.

This study examines how perceived geotechnical risk, trust in technical assistance, and willingness to pay for minimal geotechnical mitigation relate to one another among residents and construction-related professionals in Cuenca, Ecuador. Specifically, it addresses three research questions (RQ): (RQ1) What levels of perceived geotechnical risk, trust in technical assistance, mitigation adoption intention, and willingness to pay are reported by this population? (RQ2) How are these constructs, together with perceived behavioral control and economic constraint, associated with one another? (RQ3) Do these constructs differ by sociodemographic and professional-profile characteristics, particularly between technically trained and non-technical respondents? To address these questions, the present study reports a cross-sectional survey analyzed using non-parametric, confirmatory-factor, and structural-equation methods (Section 2), and discusses the implications of the findings for risk communication and mitigation-support policy (Section 4).

Within this scope, the study contributes original empirical evidence on a combination of constructs—perceived geotechnical risk, trust in technical assistance, and willingness to pay—that has not previously been examined jointly in the geotechnical risk literature, together with a validated, reusable measurement instrument that can be adapted to other Andean or Latin American cities facing comparable geotechnical exposure. The evidence generated is intended to inform local risk communication strategies and policy discussions on minimal geotechnical assistance programs, particularly regarding the relative roles of awareness, trust, and economic accessibility in shaping residents’ engagement with preventive geotechnical measures.

The problem addressed here sits squarely within the sustainable urban development agenda. Sustainable Development Goal 11 (Sustainable Cities and Communities) explicitly calls for reducing deaths, the number of people affected, and the direct economic losses caused by disasters, including those that are water- and geology-related, with particular attention to protecting the poor and people in vulnerable situations. In rapidly urbanizing Andean cities, geotechnical risk mitigation is therefore not only an engineering problem but a determinant of the social and economic dimensions of urban sustainability: unmitigated landslide and slope-instability hazards erode housing stock, displace low-income households onto ever more marginal land, and impose recovery costs that can outweigh the resources available to families and municipalities alike. Understanding whether residents and construction professionals perceive this risk, trust the technical actors who could help manage it, and are economically able and willing to pay for preventive assistance is accordingly a precondition for designing affordable, socially inclusive, and institutionally credible mitigation programs—the kind of demand-side evidence that sustainable disaster risk reduction policy in informally built Andean cities has so far lacked.

2. Materials and Methods

2.1. General Objective and Study Design

The general objective of this study was to assess the level of perceived geotechnical risk, trust in technical assistance, and willingness to pay for minimal geotechnical mitigation measures among residents and construction-related professionals in the city of Cuenca, Ecuador, and to examine the associations among these constructs, as well as their relationship with sociodemographic and professional-profile characteristics.

Consistent with the nature of the data collected and the analytical strategy described in Section 2.5, this study followed a non-experimental, cross-sectional design with a descriptive–correlational scope. The design is non-experimental because no variable was manipulated and no participant was assigned to a condition or group; responses reflect participants’ existing perceptions at the time of data collection. It is cross-sectional because data were collected at a single point in time through a self-administered online questionnaire, without longitudinal follow-up. Finally, the scope is descriptive–correlational because the study (a) characterizes the level of seven latent constructs—perceived geotechnical risk, severity and vulnerability perception, trust in technical assistance, perceived behavioral control, economic constraint/response cost, mitigation adoption intention, and attitudinal willingness to pay—within the sampled population, and (b) examines associations between these constructs and between them and sociodemographic/professional variables, without inferring causal relationships.

2.2. Study Setting

The study was conducted in the city of Cuenca, capital of the Azuay Province, located in the southern Andean highlands of Ecuador. Cuenca was selected as the study site because it is a documented case of active urban geotechnical instability within the Andean region: satellite-based monitoring combining GNSS, multi-temporal Differential InSAR, and geophysical surveys (electrical resistivity tomography and seismic profiling) has confirmed an ongoing slow-moving landslide affecting the University of Azuay campus, within the city itself, with cumulative ground displacements exceeding 20 cm and consequent structural damage to several buildings between 2021 and 2024 [21]. A complementary synthetic aperture radar and deep-learning-based analysis of active landslides in Cuenca similarly found that the toe of these mass movements lies within the city’s urban area, where deformation effects are most pronounced, while their crown extends into surrounding rural terrain [22]. At the provincial level, landslide susceptibility mapping for Azuay Province has further shown that a substantial share of the territory surrounding Cuenca presents conditions favorable to slope instability [3]. Together, this evidence indicates that geotechnical risk in Cuenca is not a hypothetical or purely regional concern but an actively monitored, locally documented condition, making the city a relevant setting in which to examine how residents and construction-related professionals perceive this risk and respond to it.

2.3. Sample

A total of 420 valid responses were retained for analysis. Participants were recruited through non-probabilistic (convenience) sampling, using an online self-administered questionnaire distributed among residents of Cuenca and individuals professionally or experientially related to residential construction. Distribution combined multiple complementary channels—including academic and professional networks connected to the civil engineering and construction sector, together with broader diffusion through social media and personal contacts—in order to reach both technical professionals and members of the general population across the city. Because distribution relied partly on academic and professional networks connected to the construction sector, individuals with an existing occupational or educational connection to civil engineering were plausibly more likely to be reached and to opt into the survey than members of the general population without such ties, a self-selection pathway that likely contributes to the professional-profile composition of the sample described below. Data collection took place between 15 May 2026 and 7 June 2026.

Inclusion criteria: (a) being 18 years of age or older; (b) residing in, or having direct professional involvement with, residential construction or housing in Cuenca, Ecuador; (c) voluntary agreement to participate after reviewing the informed consent statement; and (d) completion of the full questionnaire.

Exclusion criteria: responses from individuals who did not complete the informed consent step, or who did not meet the minimum age requirement. Because all 43 items of the instrument were configured as mandatory fields in the survey platform, no partially completed questionnaires were generated or submitted; consequently, no responses were excluded on the basis of item-level incompleteness.

The sample included participants with heterogeneous professional and occupational backgrounds: civil engineering and architecture professionals (n = 165, 39.3%), civil engineering students (n = 25, 6.0%), construction trade professionals such as builders, contractors, site supervisors, and foremen (n = 25, 6.0%), homeowners and members of the general population (n = 108, 25.7%), and individuals from other professional backgrounds unrelated to construction (n = 97, 23.1%). Combining these categories, 312 respondents (74.3%) reported some professional or occupational connection to construction, engineering, or architecture, while 108 respondents (25.7%) were homeowners or general-population residents without such a connection. This heterogeneity was an intended feature of the sampling strategy, as the study aimed to capture perceptions across both technical and non-technical segments of the urban population. A detailed sociodemographic characterization of the sample is presented in the Results section (Supplementary Table S1). No a priori power calculation was performed, because recruitment used non-probabilistic convenience sampling instead of a target-power-based design. A post hoc calculation indicates that N = 420 provides approximately 89% power to detect a small-to-medium correlation (r = 0.20) at the conservative, Holm-family-adjusted threshold used in the pre-specified tests (α ≈ 0.0033). Power falls to approximately 19% for a very small correlation (r = 0.10) at that same threshold, so non-significant results in Section 3 should be read with this ceiling in mind. The target population for this study is accordingly defined as adult residents of Cuenca and individuals with direct professional involvement in residential construction in the city, not the general population of Cuenca considered in isolation; this dual scope was an intentional design choice given the study’s interest in comparing technical and non-technical perspectives, and is reflected throughout in how findings are qualified.

2.4. Data Collection Instrument

Data were collected using a structured, self-administered questionnaire built on the Google Forms platform and originally administered in Spanish; all items and response categories were subsequently translated into English by the research team for the purposes of international dissemination, with a verified one-to-one correspondence between the original and translated versions (see Section 2.5).

The instrument comprised 15 main items (P1–P15), organized into three sections:

• Sociodemographic and contextual items (P1–P7). These items captured age group, highest level of education completed, main professional/occupational profile, prior participation in residential construction or remodeling, prior observation of physical damage signs in housing (e.g., cracks, settlement, dampness, wall tilting), proximity to a potentially unstable terrain condition (e.g., fill, slope, nearby ravine), and prior receipt of technical information or training related to soils, foundations, or safe construction practices.
• Seven multi-item Likert-scale constructs (P8–P14). Each construct was measured with five items rated on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree), grounded in Protection Motivation Theory (PMT) and the Theory of Planned Behavior (TPB):

  -

  PRG—Perceived Geotechnical Risk;

  -

  SVP—Severity and Vulnerability Perception;

  -

  CAT—Trust in Technical Assistance;

  -

  CCP—Perceived Behavioral Control/Self-Efficacy;

  -

  REC—Economic Constraint/Response Cost;

  -

  IMM—Mitigation Adoption Intention;

  -

  DAP—Willingness to Pay (attitudinal).

The complete mapping between these short construct labels, their constituent items (P8–P14), and the original survey question wording is provided in Supplementary Table S11. Item wording for each construct was adapted from established Protection Motivation Theory- and Theory of Planned Behavior-based measurement approaches used in prior hazard-behavior research, modified to the geotechnical risk context in Cuenca and translated into Spanish for local administration, prior to the expert review described below.

3.

One contingent valuation item (P15). Participants indicated, on a categorical ordinal scale (ranging from “50 USD or less” to “more than 150 USD,” plus two non-monetary response options—“depends on the type of problem” and “would not be willing to pay”) the amount they would be willing to pay for minimal geotechnical assistance, defined as a visual inspection, identification of risk signs, and preliminary technical recommendations. The categorical, closed-ended format of this item follows established methodological guidance for contingent valuation surveys [23].

Prior to its final implementation, the questionnaire was reviewed by two disaster risk management experts from the National Risk Management Secretariat (SNGR), who assessed the clarity, comprehensiveness, and relevance of each item with respect to the Protection Motivation Theory constructs being measured. The internal consistency of each of the seven Likert-scale constructs was assessed post hoc using Cronbach’s alpha, as reported in the Results section.

2.5. Data Cleaning and Preparation

Prior to statistical analysis, the dataset underwent a systematic cleaning and preparation process, fully scripted in R to ensure reproducibility. The main steps were as follows:

• Translation verification. Following translation of the instrument and response categories from Spanish to English, an exhaustive character-level check was performed to confirm that no untranslated text remained in any of the 420 × 43 response cells.
• Variable type conversion. Age group and educational attainment were coded as ordered categorical variables reflecting their natural ordinal structure; binary and nominal sociodemographic items were coded as unordered categorical factors; the 35 Likert-scale items were converted from labeled text responses (e.g., “4. Agree”) to their corresponding numeric value (1–5) for quantitative analysis, while the original labeled version was retained for frequency-based and graphical summaries.
• Recoding of the open-text professional profile variable. The original, free-text “main profile” item yielded 71 distinct categories, many represented by a single respondent, which precluded meaningful group-level analysis. Responses were therefore recoded into five mutually exclusive, theoretically defined categories (civil engineering/architecture professionals, civil engineering students, construction trade professionals, homeowners/general population, and other professional backgrounds) using a documented, rule-based text-classification procedure defined prior to any hypothesis testing. To prevent misclassification arising from lexical overlap between related terms (e.g., distinguishing professionals from students describing themselves as being in “civil engineering”), the classification rules were applied in a fixed hierarchical order, and the resulting grouping was cross-validated against the original free-text responses in full prior to use in any analysis.
• Construction of composite construct scores. For each of the seven Likert-scale constructs, a composite score was computed as the arithmetic mean of its five constituent items.
• Construction of a continuous willingness-to-pay proxy variable. The categorical ranges reported in item P15 were recoded into an approximate continuous variable (expressed in USD) using each range’s midpoint. The two non-monetary response categories (“depends on the type of problem” and “would not be willing to pay”) were coded as structurally missing for this specific derived variable, as they do not correspond to a numerical amount; these cases were retained in full for all other analyses and excluded only from analyses requiring a numeric monetary value.
• Missing data assessment. An exhaustive diagnostic check confirmed the absence of missing values across all 43 original items for all 420 respondents (0% item non-response), consistent with the mandatory-response configuration of the online survey instrument. Accordingly, no listwise or pairwise case deletion was applied at any stage of the analysis; the only missing values present in the analytic dataset were the structurally missing cases in the derived willingness-to-pay proxy variable described above.

2.6. Statistical Analysis

All statistical analyses were conducted in R version 4.5.2 (31 October 2025 ucrt) [24] using RStudio (2026.01.1, Build 403) as the integrated development environment. The following packages were used: readxl (v. 1.4.5), writexl (v. 1.5.4), dplyr (v. 1.2.0) [25], tidyr (v. 1.3.2), stringr (v. 1.6.0), forcats (v. 1.0.1), ggplot2 (v. 4.0.2) [26], svglite (v. 2.2.2), psych (v. 2.6.1) [27], corrplot (v. 0.95), Hmisc (v. 5.2.5) [28], car (v. 3.1.5) [29], and effectsize (v. 1.0.1) [30].

Descriptive statistics (frequencies and percentages) were computed for all categorical sociodemographic and contextual variables. For the seven composite construct scores, central tendency and dispersion were summarized using the mean, standard deviation, median, and range. Internal consistency reliability for each construct was assessed using Cronbach’s alpha, interpreted with attention to its known limitations and the recommendation that a high coefficient not be treated as sufficient evidence of unidimensionality on its own [31,32]. The distributional assumption of normality for each composite score was evaluated using the Shapiro–Wilk test [33], and homogeneity of variance across professional-profile groups was assessed using the Brown–Forsythe modification of Levene’s test (median-centered) [34], to inform the selection and interpretation of subsequent inferential procedures.

Given the ordinal nature of the underlying items and the non-normal distribution of the composite scores, non-parametric methods were adopted for all inferential analyses. Associations between the seven constructs, and between the willingness-to-pay (DAP) construct and the continuous willingness-to-pay proxy variable, were assessed using Spearman’s rank correlation coefficient [35]. Differences in construct scores between two independent groups were assessed using the Mann–Whitney U (Wilcoxon rank-sum) test [36,37], with the rank-biserial correlation reported as a measure of effect size [38]. Differences across more than two independent groups (professional profile and age group) were assessed using the Kruskal–Wallis H test [39], with epsilon-squared reported as a measure of effect size [40]; where the omnibus test was significant, pairwise post hoc comparisons were conducted using the Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple comparisons [41]. The association between professional profile and the categorical willingness-to-pay response was assessed using Pearson’s chi-square test of independence, with the significance level estimated via Monte Carlo simulation (10,000 replications) due to low expected cell frequencies in one response category.

Because multiple related hypothesis tests were conducted on the same dataset, a Holm–Bonferroni correction [42] was applied across the family of confirmatory hypothesis tests to control the family-wise Type I error rate; a Bonferroni-adjusted significance threshold was additionally reported, as a supplementary and more conservative criterion, for the exploratory correlation matrix among the seven constructs. Unless otherwise indicated, statistical significance was evaluated at α = 0.05 (two-tailed), based on the Holm-adjusted p-value. All analyses were fully scripted to ensure reproducibility, and figures were generated in Scalable Vector Graphics (SVG) format.

2.7. Confirmatory Factor Analysis and Structural Equation Modeling

To formally test whether the seven hypothesized constructs are empirically distinguishable from one another, and to move beyond the bivariate associations reported in Section 3.5, Section 3.6 and Section 3.7, a seven-factor confirmatory factor analysis (CFA) was estimated for the full measurement model, followed by a structural equation model (SEM) specifying theoretically motivated paths among the latent constructs. Both models were estimated in R (version 2026.01.1 Build 403) using the lavaan package, with the diagonally weighted least squares estimator (WLSMV), appropriate for ordinal, Likert-type indicators. Model fit was evaluated using the Comparative Fit Index (CFI) and Tucker–Lewis Index (TLI), for which values at or above 0.95 indicate good fit; the Root Mean Square Error of Approximation (RMSEA), for which values at or below 0.06 indicate good fit; and the Standardized Root Mean Square Residual (SRMR), for which values at or below 0.08 indicate good fit. Convergent validity was assessed for each construct using Average Variance Extracted (AVE ≥ 0.50 indicating adequacy) and Composite Reliability (CR ≥ 0.70 indicating adequacy); discriminant validity was assessed using the Fornell–Larcker criterion, under which the square root of a construct’s AVE should exceed its correlation with every other construct. The structural model specified mitigation adoption intention (IMM) as a function of the five antecedent constructs (PRG, SVP, CAT, CCP, REC), and willingness to pay (DAP) as a function of IMM and REC, allowing the indirect effect of REC on DAP through IMM to be estimated directly.

3. Results

This section presents the descriptive and inferential results obtained from the analysis of the N = 420 valid responses, organized according to the descriptive–correlational design and general objective stated in Section 2.1. Results are reported using the Holm-adjusted p-value as the significance criterion for the pre-specified family of confirmatory tests (Section 3.7), and the conventional and Bonferroni-adjusted thresholds for the exploratory correlation matrix (Section 3.5), consistent with the analytical strategy described in Section 2.6. To keep the main text focused on the findings most directly tied to the study’s objective, diagnostic, exhaustive, or non-significant outputs are reported narratively here and provided in full as Supplementary Materials, referenced by name throughout; the complete list of supplementary tables and figures is provided in the Supplementary Materials statement at the end of this article.

3.1. Sample Characteristics

The sample (N = 420) was predominantly composed of adults between 25 and 34 years of age (40.0%), followed by the 35–44 (16.9%), 45–54 (16.0%), 18–24 (13.8%), and 55 or older (13.3%) age groups. Educational attainment was high overall: 38.3% of respondents held an undergraduate degree and 35.7% held a graduate degree, while 15.5% reported technical/technological training or ongoing university studies, and only 1.2% reported basic education or none. By professional/occupational profile, civil engineering and architecture professionals were the largest group (39.3%), followed by homeowners/general population (25.7%), other professional backgrounds (23.1%), and, in equal proportion, civil engineering students and construction trade professionals (6.0% each). Two-thirds of respondents (66.4%) reported prior participation in residential construction or remodeling, and the large majority (83.8%) reported having previously observed physical damage signs in a home (e.g., cracks, settlement, dampness). Regarding proximity to a potentially unstable terrain condition, 45.5% reported none of the listed conditions, while the remainder distributed across permanent moisture or seepage (12.4%), a nearby ravine, river, canal, or runoff area (11.9%), a steep hillside or slope (9.0%), fill or artificially leveled ground (6.4%), and a nearby slope or ground cut (6.2%); 8.6% did not know. Finally, 47.9% of respondents reported having previously received technical information or training related to soils, foundations, or safe construction practices, while 45.7% had not. The complete sociodemographic and contextual characterization of the sample, including exact frequencies and percentages for every category, is provided as Supplementary Table S1, and the corresponding crosswalk between the 71 original free-text professional-profile responses and the five analytic categories used throughout this section is provided as Supplementary Table S2. The distribution of respondents by age group, educational attainment, and professional/occupational profile is additionally illustrated in Supplementary Figures S1–S3, respectively.

3.2. Reliability and Distributional Assumptions

Table 1. Internal consistency reliability (Cronbach’s α) by construct.

Construct	Number of Items	Cronbach’s α
PRG–Perceived Geotechnical Risk	5	0.844
SVP–Severity and Vulnerability Perception	5	0.857
CAT–Trust in Technical Assistance	5	0.877
CCP–Perceived Behavioral Control/Self-Efficacy	5	0.854
REC–Economic Constraint/Response Cost	5	0.828
IMM–Mitigation Adoption Intention	5	0.900
DAP–Willingness to Pay (attitudinal)	5	0.916

Note: α here denotes Cronbach’s alpha, a reliability coefficient, and should not be confused with the significance threshold (Type I error rate) denoted by the same symbol α elsewhere in this manuscript (e.g., Section 3.5).

presents the internal consistency reliability (Cronbach’s alpha) obtained for each of the seven Likert-scale constructs.

All seven composite scores departed significantly from a normal distribution (Shapiro–Wilk, p < 0.001 in all cases), which informed the selection of non-parametric procedures for the inferential analyses reported in Section 3.5 and Section 3.6; full test statistics are provided in Supplementary Table S3. Levene’s test further indicated unequal variances across professional-profile groups for IMM, DAP, and REC (p < 0.05 in all three cases); full test statistics are provided in Supplementary Table S4.

3.3. Descriptive Levels of the Seven Constructs

Table 2. Descriptive statistics of the seven composite construct scores (N = 420).

Construct	n	Mean	SD	Median	Min	Max
PRG	420	4.65	0.50	4.8	2.0	5
SVP	420	4.66	0.48	5.0	3.0	5
CAT	420	4.65	0.49	5.0	3.0	5
CCP	420	4.46	0.62	4.6	2.4	5
REC	420	4.38	0.59	4.4	1.8	5
IMM	420	4.66	0.51	5.0	1.6	5
DAP	420	4.51	0.59	4.8	2.0	5

Note: each construct score is the composite mean of its five constituent Likert items (1–5 scale); construct abbreviations follow the definitions given in Table 1.

presents the descriptive statistics (mean, standard deviation, median, and range) for the seven composite construct scores, each expressed on the original 1–5 Likert-type scale.

Figure 1 presents the pooled response distribution for each construct’s five items as a diverging stacked bar chart, and Figure 2 presents the distribution of the seven composite construct scores as boxplots.

3.4. Willingness to Pay for Minimal Geotechnical Assistance

Table 3. Distribution of declared willingness to pay for minimal geotechnical assistance (P15).

Category	n	%
Would not be willing to pay	2	0.5
50 USD or less	54	12.9
Between 51 and 100 USD	102	24.3
Between 101 and 150 USD	61	14.5
More than 150 USD	30	7.1
Depends on the type of problem	171	40.7

presents the distribution of the categorical response to the contingent valuation item (P15).

Among the subset of respondents who provided a monetary response (n = 247; i.e., excluding the structurally non-monetary categories “depends on the type of problem” and “would not be willing to pay”), the derived continuous willingness-to-pay proxy had a mean of 88.9 USD (SD = 47.0), a median of 75.5 USD, and a range of 25–175 USD.

Figure 3 presents the distribution of the declared willingness-to-pay categories.

3.5. Associations Among the Seven Constructs

Table 4. Spearman correlation matrix among the seven constructs (ρ; N = 420).

	PRG	SVP	CAT	CCP	REC	IMM	DAP
PRG	1.000
SVP	0.675	1.000
CAT	0.634	0.656	1.000
CCP	0.458	0.539	0.577	1.000
REC	0.377	0.431	0.460	0.395	1.000
IMM	0.609	0.678	0.689	0.626	0.489	1.000
DAP	0.512	0.538	0.565	0.587	0.428	0.629	1.000

presents the Spearman correlation matrix among the seven composite construct scores.

All 21 pairwise correlations were statistically significant at the conventional threshold (α = 0.05) and remained significant at the Bonferroni-adjusted threshold (α = 0.00238) computed for this 21-comparison matrix. The full matrix of exact p-values is provided in Supplementary Table S5.

Figure 4 presents this correlation matrix as a heatmap.

Four correlations directly relevant to the study’s general objective—between economic constraint (REC) and mitigation adoption intention (IMM), between REC and willingness to pay (DAP), and between mitigation adoption intention and its two antecedent constructs, perceived geotechnical risk (PRG) and trust in technical assistance (CAT)—were included in the confirmatory family reported in Section 3.7 and remained significant after Holm correction: REC–IMM (ρ = 0.489, p_Holm < 0.001), REC–DAP (ρ = 0.428, p_Holm < 0.001), PRG–IMM (ρ = 0.609, p_Holm < 0.001), and CAT–IMM (ρ = 0.689, p_Holm < 0.001).

The correlation between the attitudinal DAP construct and the continuous monetary willingness-to-pay proxy derived from P15 (n = 247) was ρ = 0.153 (raw p = 0.0161); this association did not remain statistically significant after Holm correction (p_Holm = 0.1176; Section 3.7).

As a sensitivity check on the influence of the sample’s professional composition, the two associations most central to the interpretation developed in Section 4—REC–IMM and CAT–IMM—were re-estimated separately for technical/professional respondents (n = 312) and for homeowners/general-population respondents (n = 108). Both associations replicated closely in each subsample (technical/professional: ρ = 0.471 and 0.686, respectively; homeowners/general population: ρ = 0.549 and 0.709, respectively; all p < 0.001), indicating that these core associations are not an artifact of the sample’s comparatively high share of technically trained respondents.

3.6. Group Comparisons

3.6.1. Prior Construction/Remodeling Experience

Mitigation adoption intention (IMM) and willingness to pay (DAP) were compared between respondents with and without prior construction or remodeling experience using the Mann–Whitney U test. The difference in DAP scores was statistically significant and remained so after Holm correction (W = 15,964.5, raw p = 0.0009, p_Holm = 0.0082, rank-biserial r = −0.188), with respondents reporting prior experience showing lower DAP scores. The difference in IMM scores was not statistically significant after correction (W = 17,713.5, raw p = 0.0677, p_Holm = 0.3285, r = −0.099).

3.6.2. Professional/Occupational Profile

Table 5. Kruskal–Wallis comparison of IMM, DAP, and REC by professional/occupational profile.

Construct	H	df	Raw p	p_Holm	ε2
IMM	20.026	4	<0.001	0.0049	0.048
DAP	30.511	4	<0.001	<0.001	0.073
REC	12.385	4	0.0147	0.1176	—

Note: ε² = rank epsilon-squared effect size (reported only for the two constructs with a significant omnibus test, per Section 2.6); “—“ indicates the comparison did not reach significance and is therefore not accompanied by an effect size estimate.

presents the results of the Kruskal–Wallis test comparing IMM, DAP, and REC scores across the five professional-profile groups.

The omnibus difference in REC scores across professional-profile groups did not remain statistically significant after Holm correction. Because the Kruskal–Wallis omnibus tests for IMM and DAP were significant, pairwise post hoc comparisons (Wilcoxon rank-sum test, Benjamini–Hochberg-adjusted) were conducted between all pairs of professional-profile groups; civil engineering/architecture professionals scored significantly higher than construction trade professionals and other professional backgrounds on IMM, and significantly higher than construction trade professionals, homeowners/general population, and other professional backgrounds on DAP. The full pairwise comparison matrices, including the exploratory REC comparisons, are provided in Supplementary Tables S8–S10.

Figure 5 presents the distribution of IMM and DAP scores by professional/occupational profile.

3.6.3. Age Group

IMM and REC scores did not differ significantly across the five age groups, either before or after correction (IMM: H(4) = 3.266, raw p = 0.5143, p_Holm = 0.5143; REC: H(4) = 8.449, raw p = 0.0765, p_Holm = 0.3285).

3.6.4. Reported Exposure to a Potentially Unstable Condition

PRG and IMM scores were compared between respondents who reported at least one potentially unstable terrain condition near their residence and those who did not (or were unsure). The raw difference in PRG scores was statistically significant but did not remain so after Holm correction (W = 24,728.0, raw p = 0.0152, p_Holm = 0.1176, rank-biserial r = 0.129); the difference in IMM scores was not statistically significant at any stage (W = 23,430.0, raw p = 0.1773, p_Holm = 0.3547, rank-biserial r = 0.070).

3.6.5. Professional Profile and Declared Willingness to Pay

A chi-square test of independence, with the p-value estimated via Monte Carlo simulation (10,000 replications) due to low expected cell frequencies in the “would not be willing to pay” category, indicated no statistically significant association between professional/occupational profile and declared willingness-to-pay category (χ²(20) = 30.551, p_MC = 0.0657; p_Holm = 0.3285). The full contingency table is provided in Supplementary Table S6.

3.7. Correction for Multiple Comparisons

A Holm correction was applied across the family of 15 pre-specified confirmatory tests reported in Section 3.5 and Section 3.6. Of the ten tests with a raw p-value below 0.05, three did not remain significant after correction: the correlation between DAP and the monetary willingness-to-pay proxy (Section 3.5), the comparison of REC across professional-profile groups (Section 3.6.2), and the comparison of PRG by reported exposure to a potentially unstable condition (Section 3.6.4). The remaining seven tests—REC–IMM, REC–DAP, PRG–IMM, and CAT–IMM (Section 3.5); DAP by construction experience (Section 3.6.1); and IMM and DAP by professional/occupational profile (Section 3.6.2)—remained statistically significant after correction. The complete table, including raw and Holm-adjusted p-values for all 15 tests, is provided in Supplementary Table S7.

3.8. Confirmatory Factor Analysis of the Seven-Construct Measurement Model

The seven-factor CFA showed a good fit to the data: CFI = 0.979, TLI = 0.976, RMSEA = 0.047 (90% CI [0.043, 0.051]), and SRMR = 0.043, all within conventional thresholds for good fit despite a statistically significant chi-square (χ² = 1038.01, df = 539, p < 0.001), which is expected given the sample size. All 35 standardized factor loadings were strong and statistically significant (range: 0.718–0.929, all p < 0.001), supporting the seven-construct structure of the instrument. Composite reliability exceeded 0.89 for every construct, and Average Variance Extracted (AVE) exceeded the 0.50 threshold for all seven constructs (range: 0.635–0.810), indicating adequate convergent validity throughout (

Table 6. Convergent and discriminant validity of the seven-construct measurement model.

Discriminant Validity	Max |r| with Other Constructs	√AVE	CR	AVE	Construct
Not met	0.878	0.829	0.917	0.688	PRG
Not met	0.891	0.849	0.928	0.720	SVP
Not met	0.891	0.866	0.937	0.750	CAT
Met	0.772	0.830	0.917	0.690	CCP
Met	0.697	0.797	0.896	0.635	REC
Met	0.857	0.891	0.951	0.794	IMM
Met	0.748	0.900	0.955	0.810	DAP

).

Applying the Fornell–Larcker criterion for discriminant validity, four constructs (CCP, DAP, IMM, and REC) showed a square root of AVE exceeding their correlation with every other construct. The remaining three constructs—PRG, SVP, and CAT—showed inter-factor correlations (0.837–0.891) that exceeded their respective square-root-AVE values, indicating that these three threat-appraisal- and trust-related constructs, while conceptually distinct and separately validated through their own item sets, are empirically highly convergent in this sample. This pattern is consistent with, and provides formal psychometric support for, the risk-aware, trust-oriented profile discussed in Section 4.1, in which perceived risk, perceived severity, and trust in technical assistance move closely together in this population.

3.9. Structural Model of Mitigation Intention and Willingness to Pay

The structural model, which specified mitigation adoption intention (IMM) as a joint function of the five antecedent constructs and willingness to pay (DAP) as a function of IMM and REC, also showed good fit (CFI = 0.976, TLI = 0.974, RMSEA = 0.049 [0.045, 0.053], SRMR = 0.045). The model explained 89.5% of the variance in IMM and 64.9% of the variance in DAP (

Table 7. Standardized structural path coefficients (SEM).

p	SE	Standardized β	Path
0.015	0.067	0.163	PRG→IMM
0.141	0.091	0.134	SVP→IMM
0.003	0.076	0.227	CAT→IMM
<0.001	0.037	0.337	CCP→IMM
<0.001	0.040	0.230	REC→IMM
<0.001	0.048	0.848	IMM→DAP
0.325	0.061	−0.060	REC→DAP (direct)
<0.001	0.049	0.245	REC→IMM→DAP (indirect)

).

Perceived behavioral control/self-efficacy (CCP) was the strongest predictor of mitigation intention (β = 0.337, p < 0.001), followed by economic constraint (REC; β = 0.230, p < 0.001), trust in technical assistance (CAT; β = 0.227, p = 0.003), and perceived geotechnical risk (PRG; β = 0.163, p = 0.015); severity and vulnerability perception (SVP) did not retain a statistically significant unique effect once the other four constructs were controlled for (β = 0.134, p = 0.141), consistent with its high bivariate correlation with PRG and CAT reported in Section 3.8.

Mitigation intention, in turn, strongly predicted willingness to pay (IMM→DAP: β = 0.848, p < 0.001), while the direct path from REC to DAP was small and not statistically significant (β = −0.060, p = 0.325). The indirect effect of REC on DAP through IMM was statistically significant and positive (standardized indirect effect = 0.245, p < 0.001), indicating that economic constraint is fully mediated through mitigation intention rather than acting as a direct, independent driver of willingness to pay. This mediation result provides a formal empirical mechanism for the “recognized barrier, not a suppressor” interpretation developed in Section 4.3: economic constraint appears to operate by increasing the perceived need to act, rather than by directly translating into individuals’ disposition to pay for mitigation.

3.10. Multivariable Models Controlling for Sociodemographic and Professional Covariates

Because professional status is correlated with education, age, construction experience, technical training, and risk exposure, the bivariate professional-profile comparisons reported in Section 3.6.2 cannot on their own determine whether observed differences reflect professional category specifically or these other characteristics. To address this, linear regression models were estimated for IMM and DAP (

Table 8. Selected standardized coefficients from multivariable regression models of IMM and DAP (reference professional-profile category: homeowners/general population).

DAP β (p)	IMM β (p)	Predictor
0.102 (<0.001)	0.085 (<0.001)	Education level (ordinal, 7 categories)
0.099 (0.135)	0.088 (0.131)	Construction experience (yes vs. no)
0.156 (0.026)	0.050 (0.417)	Received technical training (yes vs. no)
−0.018 (0.752)	−0.073 (0.134)	Exposure to unstable terrain (yes vs. no)
0.107 (0.216)	0.035 (0.644)	Civil engineering/architecture professional (vs. homeowner)
0.232 (0.082)	0.136 (0.247)	Civil engineering student (vs. homeowner)
−0.093 (0.476)	−0.240 (0.037)	Construction trade professional (vs. homeowner)

), and a multinomial logistic regression model was estimated for the categorical willingness-to-pay outcome (reference category: “depends on the type of problem”), each controlling jointly for age, education, construction experience, technical training received, reported exposure to unstable terrain, and professional-profile category (reference: homeowners/general population). Full model output is provided in Supplementary Table S12.

Once these covariates were included jointly, educational attainment was the only predictor significantly associated with both outcomes (IMM: β = 0.085, p < 0.001; DAP: β = 0.102, p < 0.001), while civil engineering/architecture professional status—the category driving the bivariate comparison reported in Section 3.6.2—was not a statistically significant independent predictor of either outcome (IMM: β = 0.035, p = 0.644; DAP: β = 0.107, p = 0.216). Having received technical training was significantly associated with higher DAP (β = 0.156, p = 0.026) but not with IMM. Construction trade professionals scored significantly lower than homeowners on IMM (β = −0.240, p = 0.037) once these covariates were controlled for. For the categorical willingness-to-pay outcome, a likelihood-ratio test indicated that adding professional-profile category did not significantly improve model fit beyond the other covariates (χ²(20) = 24.83, p = 0.208); educational attainment showed only a marginal improvement in fit (χ²(5) = 10.44, p = 0.064). The “would not be willing to pay” category (n = 2) produced numerically unstable coefficients in this model owing to its very small size and is interpreted with corresponding caution.

4. Discussion

4.1. A Risk-Aware, Trust-Oriented Profile Without an Apparent Fatalism Trap

The overall pattern observed across the seven constructs points to a population that is simultaneously risk-aware, confident in technical actors, and behaviorally disposed toward mitigation, and not one caught in the disengagement pattern described in parts of the landslide-behavior literature. This profile does not resemble the pattern observed by Mertens et al. [11] among landslide-exposed farmers in Uganda, where households most exposed to risk reported the lowest intention to adopt a protective measure because of low self-efficacy; given the substantial differences in risk context, institutional environment, and livelihood dependence between urban Cuenca and rural landslide-exposed Uganda, this comparison is offered descriptively rather than as evidence of a shared underlying behavioral mechanism. In the present sample, perceived behavioral control was itself comparatively high, which is consistent with Protection Motivation Theory’s core premise that strong coping appraisal—self-efficacy together with confidence in the recommended response—sustains protective intention even when threat appraisal is also high [8,10]. This reading is further supported by one of the earliest large-scale applications of PMT to natural hazards, which found that private precautionary behavior among flood-prone households depended jointly on threat appraisal and on the perceived efficacy and cost of self-protective measures, rather than on threat appraisal alone [43]. Recent applications of PMT integrated with the Theory of Planned Behavior in MDPI journals have reached a similar conclusion in flood- and multi-hazard contexts, consistently finding that coping-appraisal-related constructs (attitude, self-efficacy, subjective norms, perceived behavioral control) explain protective or preparedness intention more strongly than threat appraisal alone [15,16,17,18]; the present results, in which trust in technical assistance shows the strongest bivariate association with mitigation intention, align with that broader pattern rather than with a threat-dominated account of behavior. A plausible contextual explanation for the generally elevated threat-appraisal scores (perceived risk and severity) observed in this Ecuadorian sample is the national salience of recent, highly publicized landslide disasters, most notably the March 2023 Alausí landslide in Chimborazo Province, which caused dozens of fatalities and extensive infrastructural damage and has since been documented in the geotechnical literature as a case of anthropogenic factors compounding natural slope instability in the Ecuadorian Andes [44]. Although this study cannot establish a direct causal link between that event and the risk perception levels reported here, the temporal and geographic proximity of a high-mortality landslide to the survey period offers a plausible, non-causal explanation for why risk perception was uniformly high across sociodemographic and professional subgroups, rather than concentrated only among directly exposed respondents.

4.2. Trust in Technical Assistance and the Disconnect Between Attitudinal Willingness and Declared Monetary Value

Trust in technical assistance showed the strongest association with mitigation intention among all pairs of constructs examined, reinforcing a coping-appraisal-centered reading of protective motivation in this context and echoing prior evidence that response efficacy and confidence in the recommended course of action are central drivers of protective behavior in hazard settings [10,11]. This finding is also consistent with survey-based evidence from disaster management research showing that trust is frequently the strongest single predictor of citizens’ willingness to pay for risk-related services, ahead of resource variables such as income or social capital [20]. However, the present results diverge from that literature in one important respect: the attitudinal willingness-to-pay construct did not show a robust association with the concrete monetary amount that respondents declared they would pay for minimal geotechnical assistance once the family-wise correction was applied. Several non-mutually exclusive explanations are plausible. First, this pattern is consistent with a well-documented intention–behavior discrepancy in contingent valuation research, in which favorable attitudes expressed in a hypothetical or Likert-based format do not translate proportionally into a specific monetary commitment, a divergence attributed in the behavioral-economics literature to the activation of more favorable beliefs under hypothetical rather than real payment conditions [45]. Second, the fact that four in ten respondents selected “depends on the type of problem” rather than a fixed monetary range suggests that willingness to pay in this population is conditional on unspecified problem severity, which a single composite attitude score cannot capture. Third, unlike the goods valued in most prior contingent valuation studies cited in the introduction, such as pesticide-related health risk or general disaster-management services [19,20], a geotechnical inspection is a comparatively unfamiliar and unpriced service for most respondents in this sample, which plausibly increases the uncertainty, and therefore the instability, of any stated monetary figure relative to a general attitude. This difficulty in converting risk attitudes into a stable monetary value is not unique to the geotechnical domain: a stated-preference study of willingness to pay for earthquake risk reduction in Taiwan similarly needed to model respondents’ risk attitudes explicitly, using a prospect-theory framework, precisely because a direct, linear translation from risk perception to a monetary bid did not adequately describe the survey responses obtained [46]. Taken together, these results suggest that the DAP construct used here is best interpreted as a general disposition toward paying for geotechnical assistance, not as a proxy for the specific amount a respondent would actually commit to paying.

4.3. Economic Constraint as a Recognized Barrier, Not a Suppressor of Individual Intention

A central and somewhat counterintuitive finding is that perceived economic constraint correlated positively, rather than negatively, with both mitigation intention and willingness to pay. This result complicates a straightforward reading of response cost as a factor that suppresses protective intention, as might be expected from the general Protection Motivation Theory literature on response cost [8,10]. A plausible technical explanation lies in the wording of the economic constraint items themselves, which were framed predominantly in impersonal or third-person terms (for example, referring to cost as a barrier “for many families” or to “some people” prioritizing cost reduction) rather than as first-person statements about the respondent’s own financial limitations. Under this reading, the construct appears to capture a recognition of a structural or societal barrier instead of a personally experienced constraint on the respondent’s own capacity to act, which would explain why it coexists with, rather than undermines, individually high intention and willingness to pay. This distinction nuances the “willing but constrained” framing introduced earlier in this article: the evidence supports the claim that economic constraint is a salient and widely recognized barrier in this population, but it does not support a claim that this barrier is, at the individual level, actively suppressing the intention of the respondents surveyed. This interpretation is coherent with the descriptive–correlational, non-experimental scope of the study, which permits documenting an association between constructs but not a causal claim that removing the economic barrier would proportionally increase intention or declared willingness to pay in this specific population. The comparatively lower attitudinal willingness to pay observed among respondents with prior construction or remodeling experience offers a complementary, and similarly non-causal, technical explanation: hands-on exposure to construction may increase perceived self-efficacy in identifying or addressing geotechnical warning signs without necessarily requiring paid external assistance, a substitution between self-efficacy and response cost that is explicitly anticipated within the coping-appraisal component of Protection Motivation Theory [8,10,11]. A comparable, though not identical, pattern has been reported for seismic risk mitigation among urban homeowners, where self-efficacy showed only a weak association with the actual adoption of retrofit measures, a result the original authors interpreted as evidence that confidence in one’s own capacity to act does not automatically convert into a paid protective investment [47]; the present finding that experience is associated with lower, rather than higher, attitudinal willingness to pay is consistent with that broader observation that self-efficacy and paid mitigation behavior do not necessarily move together. The structural model reported in Section 3.9 lends this interpretation direct empirical support. Economic constraint’s positive association with willingness to pay operates entirely through its effect on mitigation intention (indirect effect = 0.245, p < 0.001), with no statistically significant direct path (β = −0.060, p = 0.325). Economic constraint therefore appears to heighten the perceived need to act, not to drive stated payment amounts on its own.

4.4. Differences Associated with Professional Profile

Civil engineering and architecture professionals reported higher mitigation intention and, more markedly, higher willingness to pay than homeowners, the general population, and respondents from unrelated professional backgrounds in the bivariate comparisons reported in Section 3.6.2, while construction trade professionals such as builders, contractors, and foremen did not show a parallel elevation relative to the general population. However, the multivariable models reported in Section 3.10 indicate that professional-profile category was not a statistically significant independent predictor of either construct once age, education, construction experience, technical training, and terrain exposure were controlled for jointly (civil engineering/architecture professionals vs. homeowners: β = 0.035, p = 0.644 for IMM; β = 0.107, p = 0.216 for DAP), whereas educational attainment remained a significant, robust predictor of both outcomes (p < 0.001 in both models). This pattern suggests that the professional-profile differences observed in the bivariate comparisons are better explained by educational attainment in general than by formal engineering training specifically, and this distinction is carried through the remainder of this section and the Conclusions. This result should be interpreted cautiously in light of a long-standing debate in the risk-perception literature regarding whether technical experts systematically judge risk differently from lay people: a widely cited review of nine empirical studies found limited consistent evidence that experts and lay respondents differ systematically in risk judgment once social and demographic confounds are accounted for [48]. More recent construction-specific evidence, however, suggests that the direction of this discrepancy may depend on the risk-related task itself: experimental work comparing frontline construction workers and managers found that managers identified more hazards correctly during hazard-identification tasks, while a separate field survey comparing contractor managers and subcontractor workers similarly found systematic differences in safety-related perception and behavior across hierarchical roles on site [49,50]. Consistent with that caution, the professional-profile differences identified in the present study were concentrated in the behavioral and monetary constructs (mitigation intention and willingness to pay) rather than in a directly tested difference in perceived geotechnical risk itself across professional groups, and the corresponding comparison for economic constraint did not remain significant after correction for multiple comparisons. Rather than indicating a general perceptual gap between experts and non-experts, these results are more parsimoniously read as reflecting differences in familiarity with, and confidence in, the value of a specific technical service, consistent with the coping-appraisal orientation discussed in Section 4.1 and Section 4.2. As with all associations reported in this study, this interpretation is offered as a plausible explanation of an observed association, not as evidence of a causal effect of professional training on willingness to pay.

4.5. Theoretical Contribution

This study extends Protection Motivation Theory and the Theory of Planned Behavior to a domain—geotechnical risk—in which the protective response is a paid technical service instead of a self-executed behavior. Two findings qualify the standard Protection Motivation Theory account of response cost: coping-appraisal constructs (trust, perceived behavioral control) predict mitigation intention more strongly than threat appraisal does, and economic constraint operates as a mediated enabler of intention instead of a direct suppressor of willingness to pay (Section 3.9; Table 7). Together they suggest a testable mechanism for replication beyond the geotechnical domain: recognition of a societal barrier may heighten, not reduce, the perceived need to act.

4.6. Practical Implications

For local authorities and engineering associations, the findings suggest that risk-communication campaigns need not focus primarily on raising awareness, since perceived risk and trust in technical assistance were already high across sociodemographic and professional groups in this sample; resources may instead be more productively directed toward reducing the economic cost of accessing minimal geotechnical assistance, for example through subsidized inspection programs, sliding-scale fee structures for lower-income households, or partnerships between local government and professional engineering associations to lower the effective cost of a first technical consultation. Because the professional-profile advantage in intention and willingness to pay identified in Section 3.6.2 is more consistent with general educational attainment than with formal engineering training specifically (Section 3.10), outreach materials may be more effective if designed for broad readability across educational levels instead of assuming a technically trained audience.

More broadly, these findings support a sustainability-oriented reframing of geotechnical risk mitigation in Andean cities: because economic constraint operates as a recognized, mediated barrier rather than a suppressor of individual intention, subsidized or sliding-scale inspection schemes are unlikely to be undermined by low motivation and can instead be designed to convert already-high risk awareness and institutional trust into affordable, equitable access to preventive assistance. Embedding minimal geotechnical inspection programs within municipal sustainable-development and disaster-risk-reduction planning—rather than treating them as a purely technical or post-disaster response—would align local practice with the social and economic pillars of urban sustainability, extending protection to the lower-income households that are disproportionately exposed to landslide and slope-instability hazards in this and comparable Andean urban contexts.

4.7. Limitations of the Study

Several limitations should be considered when interpreting the findings reported above. First, participants were recruited through non-probabilistic, convenience sampling within a single city, which limits the statistical generalizability of the findings beyond Cuenca and, at most, other Andean urban contexts sharing similar characteristics. Second, the cross-sectional, descriptive–correlational design permits the documentation of associations among constructs but does not support causal inference regarding, for example, whether economic constraint, trust, or professional training directly determine mitigation intention or willingness to pay. Third, willingness to pay was elicited through a categorical, hypothetical contingent-valuation item rather than an incentive-compatible payment mechanism, which is a design choice known in the literature to be susceptible to hypothetical bias [45]. Fourth, the professional-profile subgroups analyzed were of markedly unequal size (for example, civil engineering students, n = 25, versus civil engineering and architecture professionals, n = 165), which limits the statistical power available to detect smaller effects in less-represented subgroups and should temper confidence in the precise magnitude, though not necessarily the direction, of the subgroup comparisons reported. Fifth, the open upper category of the willingness-to-pay item required an assumed numerical value to construct the continuous monetary proxy used in the correlation analysis, introducing a degree of measurement uncertainty at the upper end of that variable. Sixth, the economic constraint items were worded predominantly at a societal rather than a personal level, which, as discussed in Section 4.3, limits the extent to which this study can directly test the classic Protection Motivation Theory prediction that personally experienced response cost suppresses protective intention. Finally, all constructs were measured through a single self-reported instrument administered at a single point in time, which does not capture actual, revealed protective behavior and may be subject to common-method variance affecting the magnitude of the associations reported. In addition, the confirmatory factor analysis in Section 3.8 supported the seven-construct structure with good overall fit and adequate convergent validity for every construct, but three closely related constructs—perceived geotechnical risk, severity and vulnerability perception, and trust in technical assistance—did not meet the strict Fornell–Larcker discriminant-validity criterion. These three threat-appraisal- and trust-related dimensions are theoretically distinct but empirically difficult to separate fully in this population, and should be read with that overlap in mind.

4.8. Future Research Directions

Building on the findings and limitations described above, several directions for future research are proposed. Replication with a larger and, ideally, probability-representative sample would allow the multivariable models reported in Section 3.10 to be re-estimated with greater precision, particularly for the smaller professional-profile subgroups (civil engineering students and construction trade professionals, n = 25 each) and the sparsely populated “would not be willing to pay” category, for which the present analysis had limited power. Longitudinal or panel follow-up studies of the same or comparable populations would help determine whether the high mitigation intention documented here translates into actual uptake of geotechnical inspections or mitigation measures over time. Future instruments could incorporate a personally framed response-cost subscale alongside a societally framed one, allowing researchers to disentangle recognition of a general economic barrier from a personally experienced constraint on the respondent’s own capacity to act. Extending data collection to other Andean cities with differing degrees of informal urbanization, such as Quito or Loja, would allow comparative, multi-city analysis within the multidimensional land-policy framework proposed for landslide risk reduction in Andean urban contexts [2]. Future willingness-to-pay research on geotechnical assistance would benefit from incentive-compatible or real-payment elicitation mechanisms, such as binding pledges or deposit-based experimental designs, to reduce the hypothetical bias associated with purely stated preferences [45]. Building on the structural model reported in Section 3.9, future studies with larger and more representative samples could extend this analysis to a multi-group structural model comparing path coefficients across professional-profile subgroups, or to a longitudinal structural model capable of testing whether the intention-to-willingness-to-pay pathway identified here also predicts actual, revealed uptake of geotechnical assistance over time, following recent integrated PMT–TPB modeling approaches published in MDPI journals [15,16,17,18]. Finally, qualitative follow-up work, such as semi-structured interviews with civil engineering students and construction trade professionals, could help clarify why these two subgroups did not show the same elevated willingness to pay observed among licensed civil engineering and architecture professionals, despite their occupational or academic proximity to the construction sector.

5. Conclusions

This study assessed the level of perceived geotechnical risk, trust in technical assistance, and willingness to pay for minimal geotechnical mitigation measures among residents and construction-related professionals in Cuenca, Ecuador, and examined the associations among these constructs together with their relationship to sociodemographic and professional-profile characteristics. The evidence indicates a population that is simultaneously risk-aware, confident in technical assistance, and inclined toward mitigation action, a profile that departs from the fatalism-trap pattern reported elsewhere in the landslide-behavior literature and is instead consistent with a coping-appraisal-driven account of protective motivation, in which trust in technical assistance emerged as the construct most closely tied to mitigation intention. Perceived economic constraint was found to coexist with, rather than suppress, individual mitigation intention and willingness to pay, suggesting that it functions in this population as a recognized structural barrier and not a personally limiting one; this nuance qualifies, without contradicting, the broader premise that cost remains a relevant obstacle to geotechnical risk mitigation in this context. The attitudinal disposition to pay did not translate consistently into the specific monetary amount respondents declared, indicating that willingness to pay for geotechnical assistance should be interpreted as a general orientation, not as a stable predictor of an exact payment. Professional profile was associated with mitigation intention and willingness to pay in bivariate comparisons, most notably between civil engineering and architecture professionals and other occupational groups; supplementary multivariable analysis, however, indicates this association is better attributed to educational attainment in general than to formal engineering training specifically, a distinction relevant to how the finding is applied in practice. Taken together, these conclusions support a revised understanding of the “willing but constrained” premise guiding this research: within the single-city, descriptive–correlational scope of the study, respondents in this sample reported high stated readiness to engage with geotechnical risk mitigation and expressed trust in the technical actors who could support it, while economic accessibility, rather than a deficit of awareness or trust, appears to be the dimension most in need of attention when designing minimal geotechnical assistance programs for residential contexts in Andean cities, pending confirmation in a more representative sample. Because the barrier identified here is economic and structural rather than a lack of awareness or trust, addressing it through affordable, subsidized, or sliding-scale geotechnical assistance programs represents a concrete pathway for advancing sustainable, resilience-oriented urban development in Cuenca and other Andean cities confronting comparable informal-construction and landslide-exposure conditions.