Empowered to Detect: How Vigilance and Financial Literacy Shield Us from the Rising Tide of Financial Frauds

Abstract

According to the literature, the advancement of information and communication technology (ICT) has increased individual exposure to scams, turning fraud victimization into a significant concern. While prior research has primarily focused on socio-demographic predictors of fraud victimization, this study adopts a behavioral perspective that is grounded in the Signal Detection Theory (SDT) to investigate the likelihood determinants of individuals becoming fraud victims. Using survey data of 671 Indonesian respondents analyzed with the Partial Least Squares Structural Equation Modeling (PLS-SEM), we explored the roles of vigilance and financial literacy in moderating the relationship between fraud exposure and victimization. Our findings substantiate the notion that higher exposure to fraudulent activity significantly increases the likelihood of victimization. The results also show that vigilance negatively moderates the relationship between fraud exposure and fraud victimization, suggesting that individuals with higher vigilance are better at identifying scams, thereby decreasing their likelihood of becoming fraud victims. Furthermore, financial literacy is positively related to vigilance, indicating that financially literate individuals are more aware of potential scams. However, the predictive power of financial literacy on vigilance is relatively low. Hence, while literacy helps a person sharpen their indicators for detecting fraud, psychological, behavioral, and contextual factors may also affect their vigilance and decision-making.

1. Introduction

The financial services sector has undergone significant transformation due to advancements in information and communication technology (ICT). Not only do those changes reshape the behavior of both service providers and consumers (Junarsin et al., 2023; Lu et al., 2024; Phan et al., 2020; Tseng & Guo, 2022), they also enable the emergence of new types of fraudulent schemes (Jewkes & Yar, 2013). “State-of-the-art” fraudulent schemes have raised serious concerns about consumer protection, as they increase the likelihood of individuals becoming fraud victims (Iman et al., 2023). Over the past decade, fraud victimization has kindled scholars’ and regulators’ attention due to its detrimental impacts on consumer well-being and financial system stability. On the one hand, fraud and scams pose a threat to a victim’s economic well-being, physical and mental health, and social relationships (Beals et al., 2015; Brenner et al., 2020; Morgan, 2021). On the other hand, fraudulent activities also endanger financial system stability by eroding consumer trust, given that the financial services industry is a trust-incentive industry (Brenner et al., 2020; Guiso et al., 2008). Globally, previous research indicates that millions of individuals have fallen victim to fraud and scams, with annual losses estimated to be $ 40–50 billion (Deevy et al., 2012; DeLiema et al., 2020). In the Indonesian context, the Indonesian Financial Services Authority (OJK) reported a total loss of Rp 139.67 trillion between 2017 and 2023 due to investment scam activities alone, including Ponzi schemes, robo-trading, and various forms of investment scams.

Although fraud victimization has emerged as a critical issue and has garnered significant scholarly attention, the existing body of literature on this topic remains limited. Most of the prior studies predominantly adopted the vulnerability hypothesis framework, which posits that individual vulnerability is the primary determinant of fraud victimization. Within this framework, researchers have focused on examining socio-demographic attributes (e.g., age, gender, education level, social affiliation, economic status, and geographic location) to investigate the likelihood of individuals becoming fraud and scam victims (Beach et al., 2016; Judges et al., 2017; Raval, 2021; Shang et al., 2022; Xing et al., 2020). However, their findings remain inconclusive and contradictory. For instance, Beach et al. (2016) found that older individuals are at a higher risk of fraud victimization. Conversely, Xing et al. (2020) provided evidence that younger people are actually more susceptible to fraud and scams because of their higher level of engagement with technology.

While the vulnerability hypothesis emphasizes the roles of demographic attributes such as age, gender, and socioeconomic status, it often overlooks the dynamic and situational aspects of human behavior that crucially affect fraud susceptibility (Deevy et al., 2012). In contrast to the vulnerability perspective, the decision-making framework highlights the dynamic cognitive and affective processes that shape individual choices under risk. Research on decision-making indicates that individual differences in cognitive style during information processing can affect susceptibility to heuristic biases, with more rational individuals demonstrating a greater ability to recognize potential risks compared to those with more impulsive tendencies (Stanovich et al., 2004). Additionally, intense emotions such as fear, greed, or despair can disturb rational thinking and lead to poor decisions (Loewenstein et al., 2001). Accordingly, two individuals with similar demographic backgrounds may respond differently toward the same stimulus depending on their emotional states at the time.

The previous findings thus suggest that examining fraud victimization that is solely predicated on socio-demographic factors is insufficient to conclude the probability of an individual becoming a fraud victim. Moreover, socio-demographic factors alone are inadequate to explain why some individuals fall victim to fraud while others do not (Van Wyk & Mason, 2001). In summary, adopting the decision-making perspective enables a more comprehensive investigation into fraud victimization by focusing on the psychological mechanisms and contextual variables that shape individual behavior. Hence, our study is purported to investigate the behavioral factors that affect the likelihood of individuals becoming fraud victims.

This study provides an array of contributions to the growing body of literature on fraud victimization. First, previous research suggests that intensive utilization of information technology, including social media, e-commerce platforms, and online shopping, induces the individual’s risk of falling victim to fraud and scams (Garg & Niliadeh, 2013; Holtfreter et al., 2010; Jansen & Leukfeldt, 2016; Reisig & Holtfreter, 2013). Employing Partial Least Squares Structural Equation Modeling (PLS-SEM), our results corroborate the notion that a greater utilization of information technology increases the likelihood of fraud victimization. Second, previous studies predominantly tested the vulnerability hypothesis and focused on socio-demographic predictors of victimization, yielding inconsistent and inconclusive findings (Beach et al., 2016; Judges et al., 2017; Raval, 2021; Shang et al., 2022; Xing et al., 2020). In contrast to their approaches, our study applies the Signal Detection Theory (SDT) (Green & Swets, 1966) to investigate the cognitive and behavioral mechanisms that may help individuals prevent fraudulent schemes. The results from a sample of 671 participants show that vigilance significantly moderates the relation between fraud exposure and victimization, indicating that individuals with a higher level of vigilance are less likely to become victims, even in high-risk fraud environments.

Third, although previous research often associates financial literacy with better fraud detection and avoidance (Judges et al., 2017; McAlvanah et al., 2015), the findings on its direct impact on victimization are mixed, with several studies reporting no significant relation or even showing a positive correlation (DeLiema et al., 2020; Drew & Cross, 2013). Our study finds a positive link between financial literacy and vigilance, confirming that individual awareness of fraudulent activity indicators increases as their financial knowledge strengthens. Nevertheless, the predictive power of this relationship is relatively low, suggesting that other factors also affect individual sensitivity to fraud and scams, as indicated by SDT. These findings help explain why greater financial literacy does not necessarily reduce the likelihood of victimization and, in some cases, could even lead to a higher probability of victimization.

This paper is organized as follows. Section 2 reviews the literature and develops hypotheses. Section 3 outlines the research methodology. Section 4 presents the findings, and Section 5 concludes the paper.

2. Literature Review

2.1. Financial Fraud and Scams

Fraudulent activities represent unethical behavior that is designed to acquire personal advantages for a perpetrator, including rights, monetary assets, or property that do not legitimately belong to them (Fletcher, 2007). The literature classifies fraudulent activities into fraud and scams. Fraud is defined as a deliberate act that is conducted by an individual or an organization to obtain illegal gains (Simborg, 2008). Financial fraud typically involves financial service providers that exploit information asymmetry to create a false sense of contract disclosure by disseminating misleading information regarding the actual characteristics of financial instruments (Black, 2006; Reurink, 2018). Such occurrences arise since financial instruments are primarily future-oriented and involve exchanging intangible rights that heavily rely on the issuer’s status and future performance (Langevoort, 1997; Lomnicka, 2008; Velikonja, 2012). Consequently, any false information provided to the markets will skew perceptions about the future potential of those rights (Reurink, 2018). Numerous fraudulent activities, including rogue trading, personal data theft, premium theft, illegal flipping, straw buying, equity skimming, air loans, and mis-selling, are classified as financial fraud (Carswell & Bachtel, 2009; Czechowska & Waliszewski, 2018; Ericson & Doyle, 2006; Fisher, 2014; Kubacki, 2022; Simborg, 2008; Wexler, 2010).

In contrast to financial fraud, a financial scam is rooted in a deceitful practice or theft strategy entirely based on lies and falsehoods (Reurink, 2018; S. P. Shapiro, 1987). In those schemes, perpetrators often adopt fake identities to gain the trust of their victims, subsequently manipulating, deceiving, or persuading them to willingly engage with the scammers, making victims voluntarily give away money, assets, or sensitive personal and financial information (D. Shapiro, 2013). The literature identifies crimes in this category as financial identity scams, such as phishing, pharming, and payment scams (Lawson, 2009; Reyns, 2018), as well as investment scams, including Ponzi schemes (Frankel, 2012; Nolasco et al., 2013). The existing literature indicates that financial fraud is frequently classified as a “low frequency, high impact event” arising from inadequate oversight within the financial services industry and/or a culture of non-compliance that ultimately results in management’s inability to identify deviant behavior among its employees (Hudson, 1998; Instefjord et al., 1998; Krawiec, 2000, 2009). Therefore, enhancing control through stringent regulations and bolstering compliance measures could mitigate financial fraud (Cole, 2023; Udayakumar et al., 2023; Zhuo et al., 2024). In contrast, financial scams are carried out by individuals who adopt false identities and often pretend to be representatives of well-known financial institutions (D. Shapiro, 2013), making them harder to curb using conventional government regulations or company policies. Hence, our study is focused on financial scams, which have proliferated significantly across various nations (Bai & Chen, 2013; Brody et al., 2007; Chryssikos et al., 2008).

2.2. Hypotheses Development

To comprehend the connection between fraud exposure, vigilance, and victimization, it is essential to understand the mechanisms behind scam activities. As discussed in the previous section, financial scams can be broadly classified into two main types: (1) identity theft and (2) investment scams. Identity theft involves acquiring confidential information such as bank account numbers, access codes for banking and computer systems, identity verification numbers, and maternal maiden names. This illicit information is then used for personal benefits, including, but not limited to, procuring loans, leasing residential properties, and conducting a multitude of other transactions under the names of unsuspecting individuals (Brody et al., 2007). Phishing and pharming are predominant techniques used to acquire personal identities illegally from unsuspecting victims. Phishing is perpetrated through disseminating e-mails to many users within a digital domain and targeting those who may be less vigilant (Brody et al., 2007). For instance, a perpetrator may dispatch thousands, or even millions, of e-mails that appear to come from a legitimate financial institution. Those e-mails typically contain privacy notifications that caution individuals of a serious issue necessitating immediate compliance with specific instructions outlined in the messages (Brody et al., 2007). Pharming is a more sophisticated variation of phishing, i.e., utilizing malicious software or programs clandestinely installed on a victim’s device. Usually, the user unintentionally downloads a hidden app that redirects the web browser to a fake website. On this fraudulent site, the perpetrator has embedded a keylogger, enabling the monitoring of keystrokes used on the legitimate website to capture login credentials (Hicks, 2005). In this scheme, victims often remain unaware that they are interacting with fake websites, as their computer screens continue to display familiar interfaces and web addresses of the seemingly genuine sites (Brody et al., 2007).

Meanwhile, an investment scam is a fraudulent scheme designed to deceive individuals into allocating financial resources toward purported investment opportunities. Those opportunities may include companies, investment funds, or projects that do not exist or may not last for the expected duration (Reurink, 2018). Such fraudulent investments often appear to be stocks, bonds, or other securities that claim to be backed by technological advancement or attractive business ventures (Blanton, 2012). Perpetrators frequently employ Ponzi schemes that masquerade as participation in innovative investment projects. The schemes often promise high returns with low risk across various sectors, including property, insurance, and precious metal mining (Frankel, 2012; Friedrichs, 2010; Geis, 2011; Naylor, 2007). Although both financial scams employ different methods and schemes, they share a common activity pattern. In this case, the perpetrators must establish a connection or approach potential victims to either excite their hopes by offering high returns with low risk or instill fear by threatening problems if they do not follow the perpetrators’ instructions (Langenderfer & Shimp, 2001). These circumstances show that the perpetrators must first gain access and establish a connection with potential victims to lure them into fraudulent schemes. However, ICT advancement has exponentially expanded a perpetrator’s reach to potential victims, making it easier for him/her to carry out internet scams, identity thefts, and Ponzi schemes (Ozili, 2020).

A string of literature documents that the level of internet utilization is associated with exposure to fraud. Holtfreter et al. (2005) indicated that the higher use of ICT, such as social media and e-commerce, increases the risk of falling victim to e-mail, website, and platform scams. Corresponding with this perspective, a recent study reveals that online shoppers have been facing various challenges, including financial fraud, privacy issues, poor product and service quality, and fake promotions (Waqas et al., 2023). A body of literature on social media shows that the ease of creating fake profiles and the high interest in these platforms have led to a significant rise in fraudulent activities. These activities encompass various tactics, including investment scams designed to establish user trust and personal information theft through phishing attacks (Bokolo & Liu, 2024). Furthermore, several studies illuminate the darker aspects of social media, where fake profiles, referred to as a “digital wolf in sheep’s clothing,” capitalize on the anonymity afforded by these platforms to perpetrate financial fraud through intricate schemes and identity theft (Dwivedi et al., 2018; Sahoo & Gupta, 2020). In addition, the phenomenon of “social bots” has emerged, manifesting as fake profiles that replicate human behavior and are operated by automated programs (Tiwari, 2017). These bots are employed to disseminate spam (Abinaya et al., 2020; Zhang et al., 2012), distribute phishing links (Shafahi et al., 2016), and engage in fraudulent activities aimed at generating profit by leveraging the popularity of their social media accounts (Boshmaf et al., 2011).

Such circumstances suggest that individuals with higher levels of information technology utilization are more susceptible to financial scams. Recent research further supports this notion, revealing that older adults, individuals with lower levels of education, and those residing in rural areas tend to be less vulnerable to financial scams due to the technological barriers they encounter in utilizing information technology (Hauk et al., 2018). Therefore, we argue that individuals who engage more intensively with the internet face a greater exposure to fraud, thereby increasing their likelihood of becoming fraud victims.

Hypothesis 1.

Fraud exposure increases the likelihood of fraud victimization.

Prior research has extensively examined the risk factors that determine an individual’s likelihood of becoming a financial scam victim. Extant literature predominantly focuses on a single hypothesis, i.e., disadvantaged and vulnerable individuals are at a higher risk of becoming fraud victims (Raval, 2021). Accordingly, researchers have focused on examining socio-demographic factors associated with vulnerability, including age, gender, education level, marital status, and income level (Beach et al., 2016; Xing et al., 2020). However, the same risk factors might lead to a different decision (Fan & Yu, 2022), suggesting that individual exposure to risk, socio-demographic characteristics, and personality traits may not fully account for the complexity of the decision-making process when faced with risk or susceptibility to financial fraud.

Signal Detection Theory (SDT) provides a methodological framework for explaining an individual’s decision-making process under uncertainty (Green & Swets, 1966; Hautus et al., 2021). SDT highlights the importance of distinguishing between signals representing key stimuli and noise, including irrelevant information (Green & Swets, 1966). The effectiveness of a decision is significantly influenced by individual sensitivity and criterion setting (Skinner & Giesbrecht, 2025). Sensitivity refers to an individual’s overall ability to distinguish between various circumstances, such as memory sensitivity in recognizing whether a particular stimulus has been previously encountered. Numerous internal and external factors, including size, duration, and visual acuity in visual tasks, shape this sensitivity (Maniscalco et al., 2024). On the other hand, criterion setting relates to individual decision-making indicators for internal interpreting and evaluating potentially ambiguous evidence to transform it into a definitive classification. For instance, a person decides to report that a visual event occurs even when there are uncertainties about it. Criterion setting primarily resides with an observer and can differ based on the specific requirements of a task (Maniscalco et al., 2024).

Studies on fraud victimization have highlighted that fraud exposure often begins when fraudsters target individuals who lack vigilance (Brody et al., 2007) and attempt to exploit the weaknesses inherent in their potential victims (Wolfe & Hermanson, 2004). Nevertheless, individuals must engage in a decision-making process that ultimately determines the likelihood of victimization following their exposure to fraud (Deevy et al., 2012). Taking the SDT lens, the likelihood of individuals falling victim to frauds and scams is greatly affected by their sensitivity to fraudulent offers and the criteria used to distinguish them. The SDT framework can be meaningfully applied to studying fraud victimization, where individuals must distinguish between fraudulent and legitimate information in complex and often ambiguous situations. From this perspective, vigilance can be seen as a determinant of sensitivity, whereas financial literacy might be a key factor influencing criterion setting.

Research on the taxonomy of vigilance indicates that vigilance is closely intertwined with an individual’s sustained sensitivity to stimuli over time, especially in tasks that involve detecting infrequent or ambiguous signals (Davies & Parasuraman, 1977; See et al., 1995). This perspective aligns with the concept of sensitivity in SDT, which describes a person’s ability to distinguish between relevant signals, such as fraudulent stimuli, and background noise, such as legitimate offers or non-deceptive contents (Green & Swets, 1966). Vigilance might play a vital role in moderating the relationship between fraud exposure and fraud victimization, as it affects how well individuals interpret and respond to deceptive cues. Although exposure to fraud is a necessary condition for victimization, it alone does not guarantee that victimization will occur, since many individuals who encounter fraudulent activities do not necessarily fall victim to it. Vigilant individuals are more likely to notice subtle inconsistencies, such as mismatched e-mail domains, linguistic irregularities, or unrealistic financial offers, which are often embedded features of scams (Davies & Parasuraman, 1977; See et al., 1995). These individuals can interrupt the pathway from exposure to fraud victimization by identifying cues of fraudulent offers or information, thereby reducing the likelihood of victimization (Brody et al., 2007). In contrast, individuals with lower vigilance may fail to recognize such cues and are more likely to be deceived.

Furthermore, financial literacy serves as a fundamental element in criterion setting, which is the second primary mechanism in SDT. Criterion setting refers to an internal decision threshold or indicator used to determine whether an ambiguous stimulus indicates a signal (e.g., fraud) or noise (e.g., legitimate information). Financially literate individuals are more likely to apply stringent and more informed evaluative criteria, allowing them to effectively filter out unrealistic or high-risk financial offers (Kersting et al., 2015; Mishra, 2019). Additionally, financial literacy can also enhance vigilance, as a deeper understanding of financial concepts and fraud indicators provides individuals with sharper frameworks to more easily identify fraudulent offers and information (Judges et al., 2017; McAlvanah et al., 2015). For instance, someone experienced in investment risk-and-return profiles is more likely to remain skeptical and attentive when evaluating unrealistic financial offer propositions, thereby fostering their sensitivity to deception. Such nuance suggests that although financial literacy may not directly increase sensitivity, it significantly sharpens the evaluative threshold, which, in turn, affects the likelihood of making correct or incorrect judgments when faced with potentially fraudulent information. Accordingly, our study conjectures that vigilance negatively moderates the relationship between fraud exposure and fraud victimization, where a higher level of financial literacy enhances individual vigilance in detecting fraudulent offers.

Hypothesis 2.

Vigilance has a negative moderating effect on the relationship between fraud exposure and fraud victimization.

Hypothesis 3.

Financial literacy improves vigilance.

The conceptual model developed in this study is presented in Figure 1.

3. Methodology

We adopted Signal Detection Theory (SDT) to develop testable and verifiable hypotheses (Zeng et al., 2021). A quantitative approach was employed, relying on primary data as the main source of empirical observation.

3.1. Instrument Measurements

This study focuses on fraud victimization as the salient area of interest. Fraud victimization is defined as an experience of being misled into providing monetary assets, personal information, or other valuable resources to a perpetrator (Button & Cross, 2017; Levi, 2017). Prior studies have highlighted two key dimensions of fraud victimization: (1) disclosure and (2) consequences. Disclosure involves the unintentional release of sensitive personal or financial data to perpetrators (Button & Cross, 2017), whereas consequences refer to financial losses and emotional distress caused by the fraudulent scheme (Levi, 2017). Fraud victimization is measured using four items, adapted from DeLiema et al. (2018) and Junger et al. (2023), which capture disclosure and consequences dimensions, as presented in

Table 1. Operationalized constructs and measurement items.

Main Construct	Reflective	Instruments
Fraud Victimization	FV1	How often did you invest your money because someone had promised high or guaranteed returns?
	FV2	How often did you pay to receive a gift, grant, inheritance, or lottery winning that you never received?
	FV3	How often did you give or lend money to someone pretending to be your relative, friend, or acquaintance?
	FV4	When responding to an e-mail or website, how often did you provide your username, password, or debit or credit card information to a stranger/outsider?
Fraud Exposure	FE1	How often did you receive notification of a lottery win or competition win, but were later informed that you had to pay a certain amount of money or purchase a product to claim your prize?
	FE2	How often have you been contacted by someone pretending to come from a legitimate organization, and asked to provide (or confirm) your personal information?
	FE3	How often did you access a website, was told you had a computer or internet problem, and then asked about personal information to resolve it?
	FE4	How often have you been promised that you would receive goods, services, discounts, or important investment profit if you transferred or invested money?
Vigilance	VL1	You are suspicious of letters or e-mails that contain spelling and grammatical errors.
	VL2	You avoid clicking links in e-mails or text messages unless you know the senders.
	VL3	You are suspicious when people you do not know contact you directly, via telephone, e-mail, social media, etc.
	VL4	You always check a vendor’s credibility.
Financial Literacy	FL1	Net worth is …… The difference between expenditures and income The difference between liabilities and assets The difference between cash inflow and outflow The difference between borrowing and saving None of the above
	FL2	Which account usually pays the HIGHEST interest? Certificate of deposit (CD) Savings account Checking account Money market account I do not know
	FL3	What are the MOST important factors that lenders use when deciding as to whether to approve loans? Marital status and the number of children Education and occupation Age and gender Bill-paying record and income I do not know
	FL4	If a consumer fails to pay personal debt, the creditor is allowed to do all of the following EXCEPT? Discuss the consumer’s debt with his/her employer Press charges against the consumer Tell a credit bureau that the account is delinquent Turn his/her account over to a professional debt collector, I do not know
	FL5	What does a credit bureau do? Approves credit applications Informs applicants on the reasons for credit rejection Extends credits to qualified applicants Provides creditors with reports of consumers’ bill-paying records I do not know
	FL6	If a credit card account has a balance carried over from the previous month, when will interest charges usually begin on a new credit purchase? On the day of the purchase One month after the date of the purchase After a two-week grace period After a two-month grace period I do not know
	FL7	Which of the following investment combinations is the most risky? A mutual fund containing 80% stocks and 20% bonds A mutual fund containing 80% bonds and 20% stocks An index fund (like the S&P500) Stock in a single company I do not know
	FL8	Many people put aside money to take care of unexpected expenses. If you have money to put aside for emergencies, in which of the following forms will it be of LEAST benefit if you need it right away? Savings account A house Stocks Checking account I do not know
	FL9	The main reason to purchase insurance is to Protect you from a loss recently incurred Provide you with excellent investment returns Protect you from sustaining a catastrophic loss Protect you from small incidental losses Improve your standard of living by filing fraudulent claims I do not know
	FL10	Rob and Molly are of the same age. At age 25, Rob began saving $2000 a year for 10 years and then stopped at age 35. At age 35, Molly realized that she needed money for retirement and started saving $2000 per year for 30 years and then stopped at age 65. Now they are both 65 years old. Who has the most money in his or her retirement account (assuming both investments had the same interest rate)? Molly, because she saved more money overall Rob, because his money has grown for a longer period of time They would each have about the same amount Unable to determine with the information provided

. Fraud exposure refers to the risk that a person or an entity might be targeted by a fraudulent activity (R. Anderson et al., 2013; Gordon & Ford, 2006). It includes various conditions and factors that increase an individual’s likelihood of encountering fraud attempts. These factors can involve the individual’s engagement with specific financial activities and utilization of digital services. Four questions from Bijwaard (2020) are adept at measuring fraud exposure, as outlined in Table 1.

Vigilance is a multifaceted concept that indicates a person’s capability to maintain focused attention to identify subtle or rare threats and to behave cautiously over time, especially in situations involving danger or deception. Vigilance is closely related to sensitivity in SDT, which involves sustained attention, sensitivity to signals, behavioral caution, and metacognitive monitoring. Individuals with high vigilance are more effective at detecting threats (or hits) and reducing false alarms, demonstrating a high ability to differentiate authentic from misleading cues (Shaw et al., 2010; Warm et al., 2018). This sensitivity is both cognitive and behavioral, as vigilant individuals commonly take preventive measures, such as confirming information and avoiding impulsive reactions (Parasuraman, 1986). Vigilance is evaluated based on responses to four questions adapted from Bijwaard (2020), as detailed in Table 1. Each question uses a five-point Likert scale for responses, ranging from “Never” (1) to “Very Often” (5).

Financial literacy is a multidimensional construct comprised of (1) awareness, (2) skills, (3) knowledge, and (4) behavior. Each contributes to an individual’s capacity to make informed financial decisions. Awareness involves recognizing financial concepts and understanding the importance of making financial decisions (Atkinson & Messy, 2012). Skill refers to the practical ability to apply that knowledge in real-world practices, such as budgeting and debt management (Lusardi et al., 2010). Knowledge encompasses understanding core financial principles such as interest rates, inflation, and portfolios, serving as the foundational cognitive basis for decision-making (Huston, 2010). Behavior, as the outcome dimension, reflects the actual financial actions individuals carry out, such as saving, planning, and responsible borrowing, demonstrating how financial literacy is put into practice (OECD/INFE, 2018). Together, these dimensions provide a comprehensive framework for assessing an individual’s financial capability, establishing criteria from the SDT perspective. In this study, financial literacy is gauged using ten questions adapted from Jorgensen (2007), as detailed in Table 1. Each correct response receives a score of 1, and an incorrect answer receives 0, resulting in a total possible score of 10. Since financial literacy encompasses multiple distinct dimensions, it is conceptualized as a formative construct in our model. These components do not represent a single underlying latent trait; instead, they work independently and together to define the constructs.

3.2. Sample Selection and Data Gathering

This study utilizes the snowball sampling technique, which streamlines the survey process by starting with a specific group of respondents and expanding to include new participants who meet the criteria based on referrals from those already surveyed. The respondent criteria are defined as follows: (1) an individual aged between 17 and 65 years old; (2) an active user of the internet, social media, or digital platforms; and (3) he/she has encountered fraudulent information or offers. The selected respondents are individuals between 17 and 65 years old for several reasons. Those under 17 are generally considered underage or less mature to effectively distinguish between appropriate and inappropriate behavior or to make independent decisions. On the other hand, individuals over 65 may experience a significant decline in cognitive function, which can impact their decision-making and the accuracy of their assessments. We also focused on participants who actively use the internet, social media, or digital platforms and who have encountered fraudulent offers. These criteria are crucial since we address issues related to digital financial fraud. Participants are expected to have well-informed perspectives on the topics being examined.

An online survey was conducted to collect primary data by distributing a structured questionnaire (Table 1) to respondents via e-mails, social networks, and messaging platforms. The questionnaire was distributed to 1000 respondents, yielding 833 completed responses, a response rate of 83%. After excluding incomplete answers, we obtained 671 clean and usable data entries.

Table 2. Demographic characteristics of respondents.

Demographic	Characteristics	Frequency	Percentage
Gender	Male	253	38
	Female	418	62
Age	17–25	302	45
	26–35	141	21
	36–45	101	15
	46–55	54	8
	56–65	73	11
Level of education	Associate degree	74	11
	Bachelor’s degree	456	68
	Master’s degree	101	15
	Doctoral degree	40	6
Internet usage (weekly)	Below 10 h	60	9
	11–15 h	362	54
	16–25 h	228	34
	Above 25 h	21	3

provides an overview of the demographic characteristics of the respondents. These individuals possess sufficient educational backgrounds and are active internet users, as indicated in Table 2. Consequently, the respondents are deemed sufficient and proper. Our analysis employed PLS-SEM to deepen the theoretical understanding of factors that influence an individual likelihood of becoming fraud victims, as postulated by Saunders et al. (2019). PLS-SEM is an appropriate method due to its ability to model complex relationships among latent constructs, such as vigilance, financial literacy, fraud exposure, and victimization, which are central to the application of the SDT framework. Conceptually, PLS-SEM aligns with this study’s objective of exploring and predicting behavioral outcomes rather than confirming established theories. Methodologically, it offers flexibility for modeling moderating effects, managing small to moderate sample sizes, and handling non-normal data distributions, which are frequently encountered in behavioral research (J. F. J. Hair et al., 2021a; Sarstedt et al., 2017). Furthermore, PLS-SEM enables the simultaneous estimation of measurement and structural models, making it especially useful for testing models that include interaction terms and complex decision-making processes.

4. Results and Discussion

4.1. Validity and Reliability of the Model

Table 3. Construct validity and reliability.

Construct	Item	Outer Loading	Cronbach’s Alpha	ρA	Composite Reliability	AVE	VIF
Fraud Victimization	FV1	0.881	0.877	0.955	0.913	0.726	3.140
	FV2	0.858					2.131
	FV3	0.745					1.749
	FV4	0.914					2.718
Fraud Exposure	FE1	0.701	0.780	0.787	0.854	0.595	1.248
	FE2	0.755					2.109
	FE3	0.834					2.169
	FE4	0.792					2.607
Vigilance	VL1	0.835	0.850	0.928	0.895	0.682	2.698
	VL2	0.828					2.785
	VL3	0.821					1.720
	VL4	0.818					1.607

exhibits the model’s construct validity and reliability. This study employs three reflective latent constructs: (1) fraud victimization, (2) fraud exposure, and (3) vigilance. Each indicator of reflective constructs in the model must meet the validity criteria for all indicators in the latent constructs calculated in the final model. J. F. J. Hair et al. (2014) suggested that a loading factor higher than 0.7 indicates a valid indicator. As shown in Table 3, the loading factors of the indicators for reflective constructs meet the validity criteria, showing that the measurement items in this study are valid.

Several well-established guidelines from the literature are important for evaluating the measurement of reflective constructs in the model. First, Cronbach’s alpha and ρA for each construct consistently exceeded the threshold of 0.7, reflecting a strong level of internal consistency across operationalized constructs (Nunnally, 1994). Second, the composite reliability for each construct ranges from 0.7 to 0.9, showing reliability for the constructs employed in this study. Third, the average variance extracted (AVE) for each construct exceeded 0.5, meaning that all constructs met the criteria for convergent validity (Sekaran & Bougie, 2016). Furthermore, the literature suggests that a model must be free from multicollinearity before establishing structural relationships (J. F. Hair et al., 2019). To confirm that our model is free from collinearity issues, the variance inflation factor (VIF) values for all predictor constructs must remain below the acceptable threshold of 5.0 (J. F. Hair et al., 2019; Henseler et al., 2016). As presented in Table 3, the highest VIF value recorded is 3.140 for FV1. This value is significantly below the threshold, indicating that the operationalized model is free from multicollinearity.

Furthermore, we also employed a formative construct, i.e., financial literacy. We follow Ali et al.’s (2018) suggestion to assess the formative construct of the model. First, we performed a redundancy analysis to evaluate the variance elucidated by the formative construct alongside the existing indicators (Cheah et al., 2018). Conducting a redundancy analysis is essential for conceptualizing all indicators associated with the formative construct that may require inclusion, given that measurement error in a formative model is recognized at the construct level (Ali et al., 2018). The redundancy analysis on our formative construct produces a path coefficient of 0.892, which surpasses the threshold of 0.707, thereby confirming the convergent validity of the construct (Cheah et al., 2018; J. F. J. Hair et al., 2021b). Second, assessing multicollinearity between formative construct indicators is a pivotal issue (Ali et al., 2018). In this case, it is crucial that the multicollinearity indicator, as measured by the VIF, remains below the established threshold of 5 (J. F. J. Hair et al., 2021a).

However, collinearity problems can also occur when VIF values exceed 3 (Ringle et al., 2015). As presented in

Table 4. Formative construct assessment.

	VIF	Outer Weight	t-Stats	p-Value
FL1	1.597	0.214	46.482	0.000 ***
FL2	1.463	0.213	45.015	0.000 ***
FL3	1.316	0.218	49.908	0.000 ***
FL4	1.094	0.130	23.092	0.000 ***
FL5	1.576	0.161	38.255	0.000 ***
FL6	1.356	0.108	20.336	0.000 ***
FL7	1.796	0.213	51.976	0.000 ***
FL8	1.264	0.214	46.132	0.000 ***
FL9	1.383	0.218	49.202	0.000 ***
FL10	1.173	0.213	44.666	0.000 ***

This table presents the test results of the formative construct operationalized in the model, i.e., financial literacy. *** on the p-values denote significance at the 1% level, respectively.

, the VIF results for the formative construct indicators are all below the threshold of 3, thus confirming that the financial literacy construct is free of a multicollinearity problem. Third, it is important to assess the significance of the weight indicators necessary to evaluate the formative construct (J. F. J. Hair et al., 2021b). In this regard, the construct indicators must demonstrate significances relative to the construct through the evaluation of the outer weights’ significances using a bias-corrected and accelerated bootstrap confidence interval (BCa) set at 95% (Aguirre-Urreta & Rönkkö, 2018; Streukens & Leroi-Werelds, 2016). All indicators of the formative construct (i.e., financial literacy) yield significant outer weights, as exhibited in Table 4. Accordingly, the criteria for the weights’ significance have been satisfied.

We used the Heterotrait–Monotrait Ratio of Correlations (HTMT) to evaluate the discriminant validity of the constructs. According to Henseler et al. (2015), a construct meets the discriminant validity criteria if the HTMT value is below 0.9. As shown in

Table 5. Discriminant validity based on Heterotrait–Monotrait Ratio of Correlations criteria.

	FE	FV	FE*VL	VL
Fraud Exposure (FE)
FE*VL	0.194
Fraud Victimization (FV)	0.256	0.154
Vigilance (VL)	0.133	0.159	0.253

This table displays the Heterotrait–Monotrait Ratio of Correlations (HTMT) results for the model. The term FE*VL denotes the interaction between the fraud exposure and vigilance constructs.

, the discriminant validity among the constructs operationalized in this study conforms to the HTMT criteria, with all values falling below the recommended threshold. This finding highlights the distinctiveness of each group, indicating that the variables used to measure them vary accordingly.

4.2. Composite Measurement Model

After establishing the validity and reliability of our model, we further evaluated its goodness of fit. The literature indicates that the estimated fit for the developed model is assessed through the standardized root mean square residual (SRMR) (Henseler et al., 2016). It has been established that an SRMR value below the threshold of 0.1 is deemed acceptable for model fitting (Aigbavboa, 2013), while values under 0.08 are regarded as adequate (Bagozzi & Yi, 2012; Henseler et al., 2014). Our analysis shows that the model’s SRMR is 0.094, which is below the standard threshold of 0.1 and is, thus, considered acceptable. We also used the Bentler–Bonnet normed fit index (NFI) to evaluate the model’s fit, as recommended by Henseler et al. (2016). A model should produce an acceptable NFI value between 0.6 and 0.9 (Henseler et al., 2016). In this study, the NFI generates a value of 0.622, which is within the acceptable threshold.

According to Ringle et al. (2015), the out-of-sample prediction procedure should be conducted using PLSPredict to assess a model’s absolute fit. This analysis provides the root mean squared error (RMSE) for both Partial Least Squares (PLS) and linear regression models (LM). The literature indicates that for a PLS model to demonstrate predictive power, its Q²predict value must be positive (Shmueli et al., 2019). Our PLSPredict analysis generates positive Q²predict values, as shown in

Table 6. Out-of-sample prediction.

	PLS		LM	PLS-LM	Remark
	RMSE	Q2 Predict	RMSE	RMSE
FV4	0.635	0.116	0.546	0.089	Moderate
FV3	0.855	0.298	0.974	−0.119
FV2	0.954	0.128	0.858	0.094
FV1	1.267	0.360	1.539	−0.272
VL4	0.606	0.279	0.646	−0.036	Low
VL3	0.963	0.067	0.897	0.036
VL2	0.885	0.016	0.805	0.036
VL1	0.835	0.031	0.788	0.036

, demonstrating our model’s predictive capability. Moreover, the RMSE for PLS must not exceed the RMSE for LM. The criteria are as follows. If all LM RMSE values are higher than those for PLS, it indicates high predictive power. If the number of indicators where the LM RMSE exceeds the PLS RMSE is equal to or greater than that where the PLS RMSE is higher than the LM RMSE, it suggests moderate predictive power. If the PLS RMSE is higher than the LM RMSE for more indicators, it indicates low predictive power. Finally, if the PLS RMSE surpasses the LM RMSE across all indicators, the model shows no predictive power (Shmueli et al., 2019). Table 6 reports that the fraud victimization construct yields two indicators where the PLS RMSE exceeds that of the LM, and two indicators where the LM RMSE surpasses that of PLS, suggesting that the operationalized model has moderate predictive power for fraud victimization. Meanwhile, our PLSPredict results for the vigilance construct discover three indicators where the RMSE for PLS exceeds that of the LM, and one indicator where the RMSE for LM exceeds that of PLS, suggesting that the model has low predictive power for vigilance. The results derived from the out-of-sample prediction procedure suggest that several additional factors may substantially affect an individual’s vigilance level.

We further evaluated the model’s predictive relevance using PLS blindfolding, which finds Q² values of 0.153 for fraud victimization and 0.082 for vigilance. While the standard guidelines suggest that an endogenous construct’s Q² value exceeds 0 to indicate predictive accuracy (J. F. Hair et al., 2019), Chin (1998) proposes that Q² values of 0.02, 0.15, and 0.35 correspond to weak, moderate, and strong predictive power. Therefore, our model has moderate predictive power for predicting fraud victimization and weak predictive power for predicting vigilance. This finding corroborates our earlier observation that an individual’s vigilance level is affected not only by the factors examined here but also by other psychological, contextual, and environmental factors not included in the current model. Therefore, the current model may provide only a partial explanation, underscoring the need for future research to incorporate a broader range of cognitive, emotional, and contextual factors to more accurately capture the multifaceted nature of vigilance. Furthermore, the model yields moderate R² values of 0.187 for fraud victimization and 0.152 for vigilance. Moreover, the global criteria for the goodness-of-fit (GoF) index indicate that a value below 0.1 is small, a value of 0.25 is moderate, and a value exceeding 0.36 is regarded as a good GoF, as outlined in Akter et al. (2011). Using the equation from Tenenhaus et al. (2005), we estimate a GoF value of 0.36, which is classified as a good model fit.

We also considered the possibility of self-reporting biases, i.e., common method variance (CMV), which is especially relevant when addressing sensitive issues such as fraud victimization. We performed an exploratory factor analysis (EFA) using principal factor extraction on 12 observed indicators across three latent variables to evaluate the presence of CMV. As shown in

Table 7. Factor loadings, eigenvalues, and uniqueness values (unrotated solutions).

Variable	Factor 1 (Fraud Victimization)	Factor 2 (Vigilance)	Factor 3 (Fraud Exposure)	Uniqueness	Eigenvalue
FV1	0.679	0.050	0.461	0.293	3.275
FV2	0.693	0.082	0.265	0.318
FV3	0.797	0.094	0.397	0.167
FV4	0.684	0.122	0.471	0.236
FE1	0.426	0.433	−0.369	0.346	1.802
FE2	0.420	0.477	−0.474	0.313
FE3	0.353	0.424	−0.519	0.342
FE4	0.459	0.391	−0.508	0.264
VL1	−0.368	0.699	0.303	0.232	2.552
VL2	−0.329	0.746	0.240	0.229
VL3	−0.391	0.646	0.251	0.313
VL4	−0.374	0.555	0.201	0.381

, the unrotated solution identifies three factors with eigenvalues greater than 1.0, which is in line with Kaiser’s criterion. The first factor accounts for 41.4% of the total variance (eigenvalue = 3.28), while the second and third factors contribute an additional 32.3% and 22.8%, respectively. The three factors collectively account for 96.5% of the variance, showcasing a strong representation of the original item variability. To further assess the CMV, Harman’s single-factor test was conducted. Since no single factor accounts for more than 50% of the variance, the findings suggest that the CMV does not pose a significant threat to data validity (Podsakoff et al., 2003). Furthermore, all items show uniqueness values below 0.40, suggesting they are well explained by the retained factors and supporting sufficient construct reliability and convergent validity (J. F. Hair et al., 2019). The clear categorization of items into logically consistent constructs provides solid initial evidence for the discriminant validity of three measured dimensions, justifying further the measurement model examination using SEM.

4.3. Structural Model Evaluation

After validating the predictive power and model fit, we analyzed the effects of the constructs on victimization likelihood. This study seeks to determine the generalizability of these effects from our sample to the broader population. Therefore, it is essential to examine the magnitude, direction, and significance of path coefficients (Ali et al., 2018; Henseler et al., 2016). To assess the statistical significance of the path coefficients, we employed the bias-correlated and accelerated (BCa) bootstrapping resampling method with 10,000 resamples, as recommended by Ali et al. (2018) and Streukens and Leroi-Werelds (2016). The results of the path coefficients are presented in

Table 8. Path coefficient results.

	Path	Coefficient	Standard Error	t-Stats	p-Value	Decision
Hypothesis 1	FE → FV	0.296	0.031	9.457	0.000 ***	Supported
	VL→FV	−0.320	0.027	11.657	0.000 ***
Hypothesis 2	FE*VL → FV	−0.240	0.035	6.811	0.000 ***	Supported
Hypothesis 3	FL → FV	0.390	0.037	10.454	0.000 ***	Supported

This table presents the structural testing results of the model. *** on the p-values indicate significance at the 1% level, respectively.

, whereas Figure 2 visually depicts the relationships between the constructs in the model. Our analysis finds that fraud exposure exhibits a positive and significant correlation with fraud victimization (β = 0.296, p < 0.05). This suggests that the likelihood of becoming a victim of financial crime increases with the increasing intensity of fraud exposure. Subsequently, vigilance is found to have a negative relation with fraud victimization (β = −0.320, p < 0.05), indicating that individuals exhibiting greater vigilance are less likely to fall victim to fraud. Furthermore, vigilance negatively moderates the relationship between fraud exposure and fraud victimization (β = −0.247, p < 0.05), suggesting that individuals with higher levels of vigilance exhibit a lower likelihood of becoming victims of financial scams, despite being exposed to intense fraudulent activity. Furthermore, we documented that financial literacy is positively related to vigilance (β = 0.390, p < 0.05). Someone’s level of vigilance increases as their knowledge of financial products, services, and markets improves. Therefore, the government and financial service providers must collaborate to enhance public financial literacy. This action can help people to avoid financial crimes, even when they are intensively exposed to fraudulent activities.

4.4. Discussion

We found evidence that individuals exposed to fraud are more likely to become scam victims, with their likelihood of being victimized increasing with greater exposure to fraudulent activities. This finding aligns with previous literature examining the fraud risk exposure process. The literature notes that the initial stage of fraud exposure begins when a perpetrator initiates communication with a potential victim through various schemes, including e-mails, text messages, or direct phone calls, with the primary goal being to steal the potential victim’s personal information or certain assets (Brody et al., 2007). These schemes are designed to arouse the hopes or fears of the potential victim (Langenderfer & Shimp, 2001). The literature also documents that perpetrators lure potential victims by offering high-yield investments with low risk in various industrial sectors, such as precious metal mining (Friedrichs, 2010; Naylor, 2007), real estate projects (Friedrichs, 2010), and insurance packages (Geis, 2011). With respect to investment prospects, perpetrators typically restructure the Ponzi scheme into specific and unique investment opportunities that offer high returns (Frankel, 2012). If tempted, the perpetrators will receive a certain amount of money or other assets intended to be invested in those projects.

Meanwhile, the perpetrators instill fear in their potential victims through privacy notices that warn individuals that severe consequences will ensue if they fail to follow the specific instructions conveyed in the notices, and that they will face even more severe consequences if they do not comply (Brody et al., 2007; Langenderfer & Shimp, 2001). This privacy information is often accompanied by fake websites or links intended to encourage users to provide their data voluntarily (Brody et al., 2007) or to inject keyloggers that allow the perpetrators to track the keys used on legitimate websites and, thus, obtain passwords (Hicks, 2005). Suppose a potential victim is lured into accessing a fake website. In that case, the perpetrator will receive information that they can harness to drain the victim’s account or sell personal information on black markets (Zeller, 2005). In executing these fraud schemes, perpetrators often use counterfeit identities to approach and build trust quickly with potential victims (D. Shapiro, 2013).

Subsequently, Hauk et al. (2018) demonstrated that individuals facing significant technological obstacles—such as residents in rural areas, older adults, and those with limited education—generally experience lower risk exposure, which reduces their likelihood of being victimized by financial fraud. This lower exposure is due to their limited engagement with digital platforms, where most modern fraud schemes prevail. In this context, fraudsters must create access routes to reach potential victims, often by developing deceptive schemes tailored to specific target groups. Historically, fraudsters acquired contact details through illegal black-market sources (Zeller, 2005). However, the rapid advancement of ICT has significantly altered this dynamic.

Contemporary fraud schemes make use of digital infrastructure to expand and enhance their targeting methods. Jewkes and Yar (2013) noted that technological innovation not only supports legitimate communication but also enables the widespread dissemination of fraudulent offers and information, allowing scammers to develop sophisticated schemes that exploit digital weaknesses. In addition, tools such as “lead list” or “mooch list”, which contain personal contact details and are readily available online, provide fraudsters with inexpensive and high-volume access to potential victims (Policastro & Payne, 2015; Shover et al., 2004). The development in fraud tactics has led to a higher level of fraud exposure as digital engagement grows (Holtfreter et al., 2005; Ozili, 2020).

From the SDT perspective, fraud victimization can be conceptualized as a perceptual decision-making challenge, where individuals need to distinguish between genuine and deceptive cues in typically ambiguous settings. In SDT terms, fraud exposure represents the signal. In this context, fraudulent offers or information, such as e-mails, texts, or phone calls, must be identified and distinguished from legitimate communications. This decision process yields two outcomes: (1) a hit (correctly detecting fraud) or (2) a false alarm (incorrectly rejecting a legitimate message). Failure consists of a miss (failing to detect an actual fraud) or a false positive (accepting a scam despite suspicion). Vigilance is essential in this perceptual framework. Vigilant individuals tend to establish higher internal thresholds for accepting information, which helps decrease the chances of misses. This corresponds with Deevy et al.’s (2012) dual-stage model of financial victimization, which posits that exposure alone does not constitute victimization. Rather, the outcome depends on the decision-making process that is shaped by personal cognitive and behavioral factors. Brody et al. (2007) and Wolfe and Hermanson (2004) also emphasized that fraudsters intentionally target less vigilant people, capitalizing on their cognitive biases and psychological weaknesses.

Furthermore, our results substantiate the conjecture that vigilance increases as knowledge about financial products, services, and markets grows. According to SDT, effective decision-making under uncertainty, such as detecting frauds, hinges on two core parameters: (1) sensitivity—the ability to accurately distinguish between legitimate and fraudulent stimuli—and (2) decision criterion, which represents a threshold for deciding if a stimulus is suspicious (Green & Swets, 1966; Skinner & Giesbrecht, 2025). From the theoretical perspective, financial literacy serves a dual purpose, i.e., spurring individual ability to detect anomalies (or sensitivity) and helping calibrate internal criteria to identify potential frauds.

In our study, financial literacy refers to an individual’s understanding of financial concepts, services, providers, responsibilities, income management, and related risks (Cossa et al., 2022). A growing body of literature suggests that individuals with greater financial literacy are more adept at recognizing early warning signs of scams (K. B. Anderson, 2019; Barthel & Lei, 2021; Engels et al., 2020; McAlvanah et al., 2015). These individuals tend to be more dubious of unusually high returns, urgent decision-making demands, or requests for sensitive information, which are common features of fraud schemes (DeLiema et al., 2019). On the contrary, financially illiterate individuals often lack reference frameworks to assess such offers, which increases their likelihood of being manipulated (Venkataraman & Venkatesan, 2018). In addition, a study by Kersting et al. (2015) documents that individuals with limited financial knowledge are less likely to understand how financial markets function, making it more difficult for them to detect suspicious or abnormal patterns in financial information.

Overall, our findings corroborate the SDT perspective. A person’s sensitivity to fraudulent cues and the decision threshold they set when assessing offers or information significantly affects the likelihood of falling victim to fraud. In this study, vigilance represents an individual’s perceptual sensitivity (a cognitive filter that helps distinguish between credible and fraudulent stimuli) while financial literacy contributes to criterion setting, enabling the development of heuristics or evaluative benchmarks to identify fraud. The interaction between these factors ultimately reduces victimization likelihood, even under a high-exposure ambiance. While fraudulent attempts have become inevitable in the digital era, individual vigilance plays a crucial role in determining the outcome of such exposure, and financial literacy helps refine the criteria used for fraud detection. Hence, preventive strategy should not only focus on minimizing exposure but also on enhancing perceptual sensitivity and decision thresholds through improved financial education.

However, it should be noted that the predictive power of financial literacy on vigilance in our model is relatively low (Q² = 0.087). This indicates that although financial literacy may help elucidate evaluative indicators, it is not the sole factor affecting vigilance or perceptual sensitivity. As posited by SDT, even well-defined criteria might yield erroneous decisions when internal psychological or external situational factors interfere with signal interpretation. Our findings shed more light on the importance of considering additional psychological, behavioral, and contextual factors that may affect vigilance. For instance, personality traits such as conscientiousness and openness have been linked to fraud vulnerability (Judges et al., 2017), while cognitive load and emotional regulation also influence decision-making quality (Pignatiello & Hickman, 2018).

Moreover, digital fatigue, cognitive decline related to aging, and past experiences with fraud could drain attention resources. A reduction in attention might lower vigilance, even for financially knowledgeable individuals (Lee & Soberon-Ferrer, 1997). Environmental situational aspects, such as time pressure or social influence, can also affect decision thresholds, increasing the likelihood of both false positives and false negatives in fraud detection (Phillips-Wren & Adya, 2020). Our findings also align with Fan and Yu (2022), showing that identical exposure to fraud risk could lead to different outcomes due to variations in perception, decision-making accuracy, and cognitive bias. These nuances might help to clarify why certain studies paradoxically report that increased financial literacy could occasionally be associated with higher victimization risk.

5. Conclusions

Fraud victimization has increasingly garnered attention over the past decade due to its adverse effects on consumers and the significant risk it poses to the efficiency and stability of financial markets. However, research in this area predominantly emphasizes socio-demographic factors when examining individual susceptibility to frauds and scams, primarily due to reliance on the theoretical framework of vulnerability hypothesis. However, existing findings on the subject remain inconclusive, underscoring the need for further research to investigate factors that determine individual susceptibility to fraud and scams (Fan & Yu, 2022). We addressed this gap in the body of literature by employing Signal Detection Theory (SDT). Utilizing Partial Least Squares Structural Equation Modeling (PLS-SEM), our analysis on 671 observational data points reveals that the individual likelihood of falling victim to fraud and scams increases with greater exposure to fraudulent activities, driven by the intensity of information technology use. Our findings substantiate previous research on the relationship between fraud exposure and fraud victimization (Garg & Niliadeh, 2013; Holtfreter et al., 2010; Jansen & Leukfeldt, 2016; Reisig & Holtfreter, 2013). Moreover, the intensity of exposure to fraud risk not only increases with the more intensive use of information technology, but it also escalates alongside the advancement in information technology (Ozili, 2020). Nevertheless, this assumption requires further empirical tests to validate its accuracy. Therefore, investigating the relationship between information technology development and fraud exposure is pivotal in examining factors that determine individual susceptibility to fraud and scams, particularly in the context of decision-making research.

Moreover, our findings confirm the SDT perspective, which posits that the quality of decision-making is heavily affected by both sensitivity and criterion setting (Skinner & Giesbrecht, 2025). In our study, sensitivity is measured as the vigilance construct, whereas criterion setting is reflected by financial literacy. We provided evidence that vigilance negatively moderates the relationship between fraud exposure and fraud victimization, implying that individuals with higher levels of vigilance are less susceptible to financial scams, even when frequently exposed to fraudulent activities. This indicates that a higher vigilance level reflects stronger individual sensitivity to fraudulent offers. In addition, the path coefficient analysis reports a positive relation between financial literacy and vigilance, confirming that higher financial literacy is associated with greater vigilance. Such circumstances arise since individuals with higher financial literacy could better comprehend the financial services industry (Kersting et al., 2015), enabling them to accurately identify key indicators that distinguish between fraudulent offers. Our overall findings suggest that governments and financial service providers broaden and enhance public financial literacy so as to help people mitigate the risk of falling victim to scams, especially when exposed to fraudulent activities.

Although both theoretical and empirical evidence support the importance of financial literacy in improving vigilance from the SDT perspective, our results indicate limited predictive power in this area. In this sense, financial literacy can strengthen awareness and decision-making by helping individuals recognize key signs of scams. However, it is not the only factor that affects vigilance. Decisions can remain impaired when various factors, such as cognitive, emotional, or contextual influences, foreshadow perceptual sensitivity. These nuances might also elaborate as to why certain studies have unexpectedly found a positive link between greater financial literacy and a higher rate of fraud victimization (DeLiema et al., 2020; Drew & Cross, 2013). Hence, a more comprehensive approach is required to understand fraud detection. Future studies should investigate how psychological and situational factors, such as trust propensity, emotional vulnerability, past fraud experiences, and decision fatigue, affect decision-making. Subsequent research may also employ longitudinal or experimental methods to observe how these factors evolve and interact with each other over time. These insights are crucial for creating more robust interventions that not only enhance financial literacy but also bolster cognitive resilience and behavioral defense against fraud.