Measuring consumer literacy about online fraud: effects on perceived financial risk and trust in online sellers

Online fraud in e-commerce represents a broad set of deceptive practices that can occur before, during, and after online transactions. This includes stolen or misused card details, suspicious offers, unauthorized financial operations, or fake online stores that imitate legitimate sellers (see the categorization by Rodrigues et al., 2022). Other authors highlight receipt fraud, shipping fraud, and non-payment fraud (Zhang et al., 2023). These fraudulent practices may occur using email communications, customer-to-customer dealings through social media, and other online transactions (Edwards et al., 2025). The broad contextualization of this important phenomenon indicates a need for domain-specific measurements that can help protect users from online fraud. The need for new measures in this context is urgent, given the serious impact of online fraud on consumers and society. For instance, in 2025, 73% of U.S. adults experienced some kind of online fraud (Pew Research Center, 2025). These figures are similar in Europe, with over 60% of respondents in a survey by the European Commission reporting serious concerns when buying online due to potential fraud (European Commission, 2025). Data from the Fraud Report in the UK shows that in 2024, scammers “stole £1.17 billion” and “70% of fraud cases started online” (UK Finance, 2025). The annual cost of online fraud globally exceeded $5 trillion in 2025.

Recent findings indicate that fraud exposure is no longer limited to traditionally vulnerable groups (e.g. less-educated, older citizens, etc.), since highly educated consumers can be victims of online scams due to their extensive engagement on the internet (Knuth and Ahrholdt, 2023). Furthermore, existing literacy constructs—such as financial, digital, information, media, and privacy literacy — are not designed to capture the specific knowledge required to identify, prevent, and respond to online fraud. Prior research shows that these literacies either omit fraud entirely or consider it within broad security dimensions, alongside ethical, technical, or health-related competencies (Alom and Ramalingappa, 2025; Li et al., 2025; Lund et al., 2025; Yu et al., 2025). In addition, empirical evidence suggests that financial and digital literacy do not consistently protect consumers from fraud and may, in some cases, increase exposure and vulnerability (Pelawi et al., 2025; Zheng et al., 2024). To fill this gap, we aim to develop a scale to measure CLOF and test some key consequences of this type of literacy on consumers' perceptions and behaviors.

The paper is structured as follows. First, we present a literature review on existing literacy measures and their relationship with online fraud, key knowledge required to identify online fraud and protect oneself from it, and the relationship of CLOF with key outcomes. The latter literature review sections are connected with the hypotheses tested in this paper. Next, we follow all steps described in the literature to develop the scale, including item generation, item reduction, purification and dimensionality, and we test its validity and reliability (Netemeyer et al., 2003). Finally, we test the nomological and discriminant validity of the scale, and present key implications as part of the conclusions.

Literacy is defined as the knowledge and set of skills that enable individuals to understand information and make effective decisions (Erdem et al., 2023). Among the literacy constructs most commonly investigated in the literature, some authors highlight the importance of financial literacy, defined as consumers' understanding of their rights, ethics regarding financial choices, as well as their ability to make complex monetary decisions (Pelawi et al., 2025; Zheng et al., 2024). Other important constructs examined in the literature include information literacy, defined as the ability to evaluate, search, understand, and use information effectively (Li et al., 2025); media literacy, the knowledge required to understand, analyze, evaluate and comprehend media messages (Erdem et al., 2023); privacy literacy, described as individuals' awareness and knowledge of institutional data practices and the ability to apply different privacy coping strategies (Lund et al., 2025); and digital literacy—the knowledge and ability to interpret digital content, assess its credibility, create new content, and communicate with appropriate tools (Ng, 2012). Table A1 of the Web Appendix provides a full view of these literacy constructs and measurements.

This review shows that CLOF is not properly captured in existing scales, and that the relationship between these literacy constructs and online fraud is far from being clear. For instance, prior studies indicate that financial literacy plays a protective role, reducing the probability of fraud-related losses or improving fraud detection (He et al., 2025), while other studies find that financially literate individuals are more likely to be exposed to fraud attempts, suggesting that greater knowledge may increase fraud victimization (Rey-Ares et al., 2024; Zheng et al., 2024). Some authors indicate that financial literacy does not necessarily prevent fraud victimization once exposure occurs, with no significant association between literacy and actual financial losses (Rey-Ares et al., 2024). Recently, Yang and Li (2025) found a non-linear relationship, where higher financial literacy reduces fraud risk at low levels but increases fraud risk once literacy becomes very high. In conclusion, existing financial literacy scales do not seem to capture and clearly show the relationship between knowledge and online fraud.

Regarding digital literacy, the majority of authors have focused on knowledge about the technical and cognitive components that allow individuals to assess and use digital information properly. However, in the last few years, some authors have included an additional dimension in their measurements, namely “security” (e.g. Alom and Ramalingappa, 2025; Li et al., 2025; Lund et al., 2025; Yu et al., 2025). All digital literacy scales that include items associated with online security tend to combine several concepts within a single dimension, including measures such as “I know the consequences of illegal downloading of music and movies”, “I understand copyright issues”, and “I follow ethical principles while using content in the digital environment” (e.g. Alom and Ramalingappa, 2025). Importantly, although some recent studies have included a security dimension in their scales, they do not include a measurement specifically associated with fraud (e.g. fraud detection, fraud victimization, etc.). Accordingly, more research is needed to understand the factors that help to prevent online fraud.

Prior literature suggests that literacy in adjacent domains (e.g. general digital literacy, financial literacy, or cybersecurity awareness) does not consistently capture consumers' ability to identify and protect themselves from online fraud. While these streams of research capture important skills, they tend to focus on broader competencies, rather than the specific knowledge required to recognize and respond to fraudulent practices in e-commerce contexts.

However, some authors have identified specific types of knowledge that function as mechanisms to prevent fraud on the internet. Particularly, some authors highlight that “… identity theft is the most cited reason for privacy worries among households” (Bian et al., 2023, p. 9). In fact, knowledge about the type of information that can be requested by companies, and the circumstances in which it should be shared, is key to preventing fraud (Balakrishnan et al., 2025; Benson et al., 2015; Bian et al., 2023). Hence, consumers need to be “… aware that malicious websites may lead to identity theft …” (Balakrishnan et al., 2025, p. 8), and that sharing personal information online can be very dangerous (Benson et al., 2015; Bian et al., 2023). This positions knowledge about identity theft online (i.e. knowing to what extent and in which circumstances a company can ask for personal/private information) as a key component of CLOF.

In addition, to identify and protect themselves from fraud, consumers should have knowledge that allows them to verify whether an online seller is legitimate (Balakrishnan et al., 2025). Consumers should be able to verify that the company exists and is genuine when acquiring a product or receiving information (Narayanan et al., 2012), because users should be “… aware that there are a lot of fake websites on the internet” (Balakrishnan et al., 2025, p. 8). Some authors suggest that “a consumer's inability to discriminate fraudulent sites is a serious obstacle for the viability of e-commerce” (Narayanan et al., 2012, p. 1615). This highlights the ability to verify the legitimacy of sellers as another key knowledge domain associated with CLOF.

In connection with the above, some authors emphasize the importance of consumers' ability to recognize red flags in messages and webpages, and react accordingly (Balakrishnan et al., 2025; Lyu et al., 2025). This includes being cautious when receiving unwanted emails, suspicious links, or poorly constructed websites. Hence, consumers should “… know which suspicious pop-up messages on websites to be ignored” (Balakrishnan et al., 2025, p. 8), and “identifying inconsistent cues is necessary for successful detection” (Lyu et al., 2025, p. 3).

Together with this, some authors emphasize the importance of using technology and protection tools. “Internet users can learn how to behave securely and identify artefacts, such as security notices …” (Benson et al., 2015, p. 37), “have antivirus software up-to-date,” “use privacy settings to block malicious websites” (Balakrishnan et al., 2025, p. 8), and rely on “security protection mechanisms …” (Benson et al., 2015, p. 37). For instance, Fonseca et al. (2022) found that users “who had antivirus software installed and updated were less likely to be victims of online consumer fraud” (p. 767). Hence, knowledge and ability to use these protection tools become key in CLOF.

Some authors also highlight consumers' awareness of the need to read terms and conditions online, something that, according to prior studies, many consumers do not properly do (Chawla and Kumar, 2022). Reading these clauses has been recognized as key to identifying “… inconceivable terms and conditions” (Chawla and Kumar, 2022, p. 590), which makes “… reading and understanding the small print” (Ross et al., 2014, p. 430) a relevant mechanism to avoid online fraud (Chawla and Kumar, 2022).

Finally, one of the most important concerns associated with online fraud is consumers' lack of knowledge about how to properly report it, which can affect both the magnitude of the problem and future protection practices (Cross, 2015; Fonseca et al., 2022; Lim and Letkiewicz, 2023). Prior research identifies “a lack of knowledge of who to report the incident to” (Cross, 2015, p. 199), as users often indicate that they “… did not know where and to whom I could report the situation” (Fonseca et al., 2022, p. 764). Consistently, Lim and Letkiewicz (2023) argued that consumers' ignorance of the reporting process constitutes one of the main problems associated with online fraud.

A complete review of these themes, including a larger number of studies and key quotations, is presented in Table A2 of the Web Appendix. Through this in-depth review, we identified six types of knowledge that have been documented in the literature. To further support the importance of these types of knowledge, the next section explores findings from adjacent literature and other institutional sources.

Besides research directly focused on online fraud, adjacent literature (e.g. cybersecurity behavior, digital resilience, and technology threat avoidance) further supports the relevance of these areas of knowledge. For instance, research on cybersecurity behavior emphasizes awareness of scams aiming to “… steal consumers' personal identity data and financial account credentials …” and “online identity theft …” (Ertan et al., 2018, p. 18), as well as the importance of fraud reporting (Afzal et al., 2024), recognizing legitimate companies (Ertan et al., 2018), and complying with security measures (Moustafa et al., 2021). Similarly, research on technological threat avoidance highlights the need to identify threats such as “online identity theft” and deceptive practices where “… the attacker creates a fraudulent website, which has the look-and-feel of a legitimate website …” (Arachchilage and Love, 2014, p. 305), including “fake replicas” (Arachchilage et al., 2016, p. 185). Research on digital resilience also discusses the “issue of identity theft” (Shandilya et al., 2024, p. 281), the increasing “… imitation of legitimate companies to trick individuals …” (Shandilya et al., 2024, p. 6), the identification of red flags (e.g. suspicious hyperlinks), and the use of protection tools such as strong passwords and updated antivirus software (see also Yan et al., 2026). Some authors also discuss the lack of awareness of reporting protocols (Kraiwanit et al., 2025), coinciding with the themes identified in the prior section.

In addition, these themes concur with data from key institutions, such as the European Commission, the Corporate Finance Institute (CFI), the Federal Trade Commission (FTC) and Financial Industry Regulatory Authority (FINRA) in the USA, Information Commissioner's Office (ICO) in the UK, and Canadian Securities Administrators (CSA) (see Table A3 of the Web Appendix). Hence, these forms of knowledge provide the foundation for the development of the present construct. Specifically, the construct encompasses (1) knowledge about identity theft, (2) verification of the legitimacy of companies, (3) Knowledge of how to report fraud, (4) recognition of red flags, (5) knowledge and use of protection tools, and (6) reading terms and conditions as a protection practice. In the next section, we extend the literature review by examining the relationship between CLOF and other consumer perceptions and online behaviors.

In the context of online buying, perceived risk is defined as the consumer's sense of uncertainty, potential loss, or harm when making purchases on the internet (Ha and Pan, 2018). Besides individual differences that lead to different users' perceptions, prior research shows that website attributes can reduce or increase consumers' perceived risk during online transactions. For instance, website quality, as well as the store image, can make consumers perceive lower risks (Kim and Lennon, 2013). In addition, perceived risk decreases when online stores use safety cues, such as privacy and security seals (Hu et al., 2010). Platform affordances can also reduce consumers' perceived risks by allowing them to inspect products, interact with sellers, and access social information before purchase. The relationship between platform affordances and users' perceived risks depends on users' capabilities to navigate the possibilities that some platforms allow (Sun et al., 2019), as well as users' knowledge to assess the quality of the website, identify the legitimate online vendors, and the veracity of seals (Kim and Lennon, 2013). Hence, besides characteristics of the website and platform affordances, perceived risk might be associated with users' knowledge.

Focusing on the connection between literacy and perceived risks, first, it is important to acknowledge that the literature presents distinct dimensions of perceived risk. For instance, some authors focus on perceived privacy risk, defined as the risk of private information being disclosed, violated, or misused (Phamthi et al., 2024). Others examine performance risk (a.k.a. Quality risk; uncertainty about whether the product will perform as advertised), financial risk (i.e. undesirable monetary loss when buying online), and security risk (i.e. potential harm to property or personal safety) (Ha and Pan, 2018). Additional dimensions of perceived risk in online buying include functional risk, price risk, time risk, social risk, service risk, delivery risk, source or information accuracy risk, and psychological risk (see full review and sources in Table A4 in the Web Appendix).

Focusing on privacy risk, some authors suggest that, because more knowledgeable consumers are more aware of potential dangers associated with information disclosure, they may become more skeptical (Cheung et al., 2015). Less knowledgeable consumers may perceive lower psychological risk precisely because they do not worry about certain potential problems when buying online (San Martín et al., 2011). A similar relationship has been suggested—although not empirically tested—for performance, delivery, and service risks (Zhang and Hou, 2017). Thus, higher knowledge may increase some types of perceived risk rather than reduce it (Rey-Ares et al., 2024; Zhang and Hou, 2017).

However, some authors propose the opposite relationship between knowledge and consumers' perceived risk (Ha and Pan, 2018; Maziriri and Chuchu, 2017). Specifically, some studies indicate that most consumers tend to seek out and rely on accurate information when they perceive that an online transaction may lead to monetary loss (i.e. financial risk). Other authors agree that, in the context of potential financial harm, lower knowledge is typically associated with higher perceived risk (Ha and Pan, 2018; Maziriri and Chuchu, 2017). Maziriri and Chuchu (2017) argued that, regarding financial risk, perceived risk decreases as consumers become more informed or experienced, and this type of risk increases when they lack knowledge.

Among other conclusions, this review shows that the relationship between consumers' knowledge (i.e. literacy) and perceived risk may vary across different types of risk examined in the literature, as well as the type of knowledge captured. In this study, we focus on a specific type of knowledge (i.e. CLOF) and a specific type of risk, namely, financial risk. We focus on this type of risk since the pursuit of others' money is one of the most common and harmful practices of online scammers (Brenner et al., 2020). Building on prior research showing that, unlike other types of risk, lower knowledge is associated with higher perceived financial risk (Ha and Pan, 2018; Maziriri and Chuchu, 2017), we propose that higher CLOF is associated with lower perceived financial risk:

In connection with the above, research indicates that any form of financial loss (e.g. credit card fraud) deters online shopping. In fact, when consumers perceive higher levels of financial risk, they are less likely to shop via the internet and the total amount spent online will also be affected (Forsythe et al., 2006). Thus, perceived financial risk is negatively associated with consumers' online buying intensity, leading to the following hypothesis:

Online fraud has important consequences besides direct victimization, as fear of fraud can decrease trust in online sellers and digital transactions (Elbeltagi and Agag, 2016; Shukla et al., 2026). Even when consumers do not experience fraud themselves, the perceived possibility of deception, identity theft, or financial loss can affect their engagement and participation in e-commerce (Narayanan et al., 2012). Therefore, examining trust in online sellers and online buying intensity helps us understand whether CLOF is linked not only to feeling more protected against financial harm, but also the relationship between this knowledge and consumers' participation in e-commerce.

Trust is defined as a person's belief that they can rely upon a promise made by another (e.g. a seller), and that the other will act with goodwill (Agag and El-Masry, 2017; Elbeltagi and Agag, 2016). Besides this general definition, prior research has defined different dimensions of consumers' trust in online contexts, including trust in the website, defined as consumers' belief that the website's technology, security, and information are reliable, making it safe to conduct transactions (Agag and El-Masry, 2017); trust in the seller (vendor or store), defined as consumers' belief that the seller is honest, competent, and will deliver products or services as promised (Pizzutti and Fernandes, 2010); trust in the reviewer/recommender before making the purchase; dispositional trust (i.e. a person's general tendency to trust others, shaped by personality and cultural background); or institution-based trust (i.e. trust that arises from the belief in the supporting structures, safeguards, and conditions for safe transactions) (Papadopoulou et al., 2001). Other types or dimensions of trust in online contexts examined in the literature include trust in online payment, trust in group members (during group-buying transactions), deterrence-based trust, and familiarity-based trust (i.e. “I know the person”) (see all definitions and sources of each type of trust examined in the literature in Table A5 of the Web Appendix).

The relationship between consumers' literacy and trust remains inconsistent in prior studies, partly due to these different dimensions of trust. For instance, literate consumers might become more skeptical and trust a webpage less, precisely due to “sufficient skills and knowledge that allow them to consider that things online may go wrong” (Agag and El-Masry, 2017, p. 349). However, several authors argue that for other types of trust, the direction of the relationship is the opposite. For instance, when it comes to trust in online payment, institution-based trust (safety measures), and trust in the accuracy of the products sold or advertised online, several authors agree that higher knowledge leads to higher trust (Papadopoulou et al., 2001; Pizzutti and Fernandes, 2010).

Thus, the direction of the relationship between literacy and trust will depend on the type of knowledge and the dimension of trust examined. In this research, we focus on how knowledge about potential online fraud is associated with trust in online sellers, which has been mentioned as one of the most important deterrents to e-transactions nowadays (Elbeltagi and Agag, 2016; Knuth and Ahrholdt, 2023; Shukla et al., 2026). In this context, it is expected that consumers will trust sellers more if they are able to identify facts that prove their goodwill and legitimacy (Walczuch and Lundgren, 2004). Other authors agree that knowledge is positively associated with trust in online sellers. For instance, Riquelme and Roman (2014) indicated that consumers with a high level of knowledge of trust assurance factors (i.e. the ability to identify safety measures) will perceive online sellers to be more trustworthy than consumers with a low level of knowledge. Therefore:

It is well established that consumers who trust sellers online more tend to make more online purchases (Riquelme and Roman, 2014; Walczuch and Lundgren, 2004). Thus, we expect that trust in sellers will be associated with online buying intensity:

Finally, we theorize that CLOF may also be directly associated with online buying intensity. Prior research shows that knowledge about common fraud practices can increase consumers' willingness to engage more frequently and extensively in e-commerce activities (Walczuch and Lundgren, 2004). Consumers who are literate about online fraud are better equipped to recognize warning signs, evaluate the legitimacy of sellers, and understand the role of payment protection, thereby reducing hesitation and uncertainty during online purchasing (Riquelme and Roman, 2014). In fact, consumers who feel competent in managing online risks are more confident in their purchase decisions and tend to buy more online (Riquelme and Roman, 2014; Walczuch and Lundgren, 2004). Therefore:

The scale development process in this research involved all the steps recommended in the literature (Netemeyer et al., 2003), and used in prior research (e.g. Agag et al., 2016; Ho et al., 2020; Su et al., 2024; Yu, 2011)—item generation, item reduction, purification and dimensionality of the scale, and validity and reliability. Given the characteristics of this scale, we considered factors associated with literacy measurements (Luna-Cortes, 2026). The different studies described next (i.e. Q-sort, pilot test, and formal studies) were conducted in different countries. These stages served different methodological purposes within the scale development process. The Q-sort was used as an expert-based item-reduction procedure, with the purpose of assessing item clarity, redundancy, and conceptual fit with the proposed construct. The selection of the sample in this step was purposeful, choosing experts who helped with the quality of the process. The pilot test was used to examine whether respondents could understand the wording of the items and whether any phrasing required refinement. These preliminary stages were not treated as evidence that results from one country automatically generalize to another. Rather, they were used to refine the content and wording of the scale before psychometric testing. The reliability and validity of the scale were then examined using samples from the USA, Sweden, and the UK The inclusion of different countries at this final stage was intentional. Since online fraud is a global phenomenon, we aimed to observe the reliability and validity of the scale in different countries.

Participants of the focus group included four scholars from Sweden (purposive sampling; three males, one female; age: 45–65). Two of the scholars worked in the field of computer science and had a background in research on cybersecurity, while the other two were senior researchers in marketing at a Swedish business school who conducted research on fraud. Experts were selected instead of general consumers for this step in the item generation because CLOF reflects technical and institutional knowledge about online fraud that benefits from expert judgment, which ensures adequate domain coverage and content validity at early stages of scale development. A standardized set of open-ended guiding questions was used to structure the discussions (see Table A6 of the Web Appendix).

As a result of this process, the following initial factors were confirmed to start the second part of the discussion: (1) Knowledge about identity theft; (2) Verification of the legitimacy of companies; (3) Knowledge of how to report fraud; (4) Checking red flags that prevent being a victim of online scams; (5) Knowledge and use of tools to protect against scams; (6) Reading terms and conditions as a protection practice. During the second part of the discussion, specific knowledge linked to each of these factors was explored. During the following week, four items were generated for each of these topics (i.e. 24 items) (Table A7 of the Web Appendix presents the full descriptions of the initial items on the scale).

To illustrate the item generation process, we provide an example related to the dimension “Verification of the legitimacy of companies”. During the focus group discussions, experts emphasized that a key indicator of fraudulent online sellers is the absence of verifiable company information, such as a physical address, registration details, or accessible customer service. We translated this knowledge into an item reflecting consumers' awareness of such verification practices, and crafted the item “When I purchase a product online, I check information about the company to verify that it exists”. The wording was deliberately kept general and self-assessed, to ensure applicability across different online shopping contexts, and to avoid measuring experience with a single platform or transaction type.

Three academics from the marketing department of a university in Spain assessed the crafted items. They were chosen because they were researchers with an extensive background in scale development and could provide feedback on the quality of the items and the structure of the scale. We asked the three participants, separately, to indicate if and why they “agree” or “disagree” that the items were valid to measure the construct, considering its fit with the term (i.e. consumer knowledge about practices and information to avoid online fraud). We considered only the items for which all experts “agreed” on their validity.

The three participants agreed that eighteen out of the twenty-four items properly measured knowledge about CLOF. One participant thought that five items were redundant, measuring ideas that had been already captured in an item before. Thus, these items were deemed unnecessary and were eliminated. Another item was eliminated since, according to one of the respondents, it was difficult to answer due to intricate phrasing, while its idea was already captured in a previous item as well. After the discussion, we decided to simplify the scale and include three items for the six potential factors, leading to eighteen items in total.

Ten students (nine females; mean age: 21) from a university in Sweden answered a questionnaire formed by the 18 items. All courses are taught in English at this university. Thus, the questionnaire was in English. After answering the questionnaire, we asked if they encountered difficulties in understanding any of the items on the scale. Based on their responses, the phrasing of two items was improved to reduce some ambiguity. No further item was deleted or incorporated at this stage. This process resulted in an 18-item scale, which was used in the following steps (see all items in Table A8 in the Web Appendix).

We conducted two studies to test the reliability of the scale. The questionnaire included the 18-item scale and demographics at the end. Study 1 A was conducted in the USA using the Prolific Research Panel, gathering 250 respondents, of whom eight were removed due to failing the attention check (convenience sampling, N = 242; 65% women; Mean age: 41.55, SD = 12.92; 67% with a university degree). We conducted Study 1 B in the city of Jönköping, Sweden. In this case, we printed and distributed questionnaires on the street (convenience sampling, N = 150; 68% women; Mean age: 23.44, SD = 3.65; 88% held a university degree or were conducting their studies at the university).

First, we conducted an exploratory factor analysis (EFA). The EFA was performed using Varimax Rotation and Principal Component Analysis on all items using IBM SPSS Statistics 23. The reliability of measurements and appropriateness of factor analysis were analyzed using Kaiser-Meyer-Olkin (KMO) and Bartlett's test of sphericity. The results showed that the KMO value was greater than 0.70, and Bartlett's test p-values were significant. Hence, the items were reliable to continue with the factor analysis. Six factors explained 82.89% of the variance. To check which items contributed to each of the six factors, a Rotated Component Matrix was generated (i.e. varimax rotation), which showed loadings higher than 0.50 in all factors (see Table 1).

Next, Confirmatory Factor Analysis (CFA) was conducted with AMOS for SPSS. The results showed that a six-dimensional factor model provided a good fit to the data in the study in the USA (χ2 = 182.06, df = 120, p < 0.01; GFI = 0.92; CFI = 0.98; TLI = 0.98; RMSEA = 0.046). As can be observed in the dimensions and the items of the scale, four factors (i.e. Knowledge about identity theft, Verification of the legitimacy of companies, Knowledge of how to report fraud, Checking red flags) are associated with knowledge-based literacy, while two factors (i.e. Use of protection tools, and Reading terms and conditions) are associated with behavioral practices. We ran a second-order model, distinguishing these two higher-order factors. The results showed lower model fit indices (e.g. χ2 = 228.08, df = 128; GFI = 0.89; CFI = 0.91; TLI = 0.92; RMSEA = 0.067). Thus, we continued the analyses with a first-order model (see Cronbach's Alpha, Composite Reliability, Average Variance Extracted results in Table 2). Focusing on the first-order model, all factor loadings were significant (p < 0.01), and standardized loadings ranged from 0.56 to 0.98 with an average of 0.84.

Because all items were measured using a self-report Likert scale within the same survey, this could raise the likelihood of common method variance (CMV). Thus, we carried out some analyses to test potential CMV. Following the established method for Harman's one-factor test, we included all items into an EFA, checking whether a single factor emerged and accounted for more than 50% of the total variance. This would suggest that CMV may be a serious concern. The test showed that the largest factor contributed 22.90% of the total variance (see Table A9 in the Web Appendix), indicating that no single factor explained the majority of the variance.

Furthermore, we used CFA to detect CMV by introducing an additional latent factor—a hypothetical variable that represents potential method bias. Precisely, we crafted the baseline model in AMOS for SPSS. In addition, we introduced a second model where all items loaded onto a single unmeasured latent factor representing CMV. Then, we compared the standardized factor loadings between the two models. We observed that the differences in standardized factor loadings were lower than 0.1 for most factors, except for three factors, which were between 0.1 and 0.2. None of the differences were higher than 0.2 (see Table A10 in the Web Appendix). These results suggested that CMV was unlikely to be an issue in the study.

Following the same criteria as in Study 1 A, EFA results showed that the KMO value was greater than 0.70, and Bartlett's test p-value was significant in this study as well. Six factors explained 79.11% of the variance. The Rotated Component Matrix showed loadings higher than 0.50 (see Table 1). The results of the CFA (AMOS-SPSS) showed that a six-dimensional factor model provided a good fit to the data for the study in Sweden (χ2 = 202.78, df = 120, p < 0.01; GFI = 0.86; CFI = 0.95; TLI = 0.94; RMSEA = 0.070). Similarly, the second-order model led to poorer model fit indices (χ2 = 219.78, df = 128, p < 0.01; GFI = 0.85; CFI = 0.90; TLI = 0.89; RMSEA = 0.079). The results showed acceptable values for CR and AVE (see Table 2). All factor loadings were significant (p < 0.01), with standardized loadings ranging from 0.53 to 0.97 and an average of 0.80. CMV was tested, which led to similar results as in Study 1 A (Harman's test—21.57% of the total variance; Common Latent Factor—differences in standardized factor loadings lower than 0.2) (see Tables A11 and A12 in the Web Appendix).

The goal of this study was to measure the relationship of CLOF with consumers' perceived financial risk when buying online (H1), and the relationship of perceived financial risk with consumers' online shopping intensity (H2). In addition, we tested the relationship of CLOF with consumers' trust in online sellers (H3), and the relationship of trust in online sellers with consumers' online buying intensity (H4). Finally, the direct relationship between CLOF and online buying intensity was tested (H5).

Four hundred participants from the UK filled out a survey on Prolific, from which 29 responses were deleted due to failing the attention check (convenience sampling, N = 371; 67% women; Mean age: 32.50, SD = 13.01; 65% with a university degree). After a brief introduction and asking for consent, the survey presented four items that measured perceived financial risk when buying online (e.g. “Purchasing products online can be a financial risk for me”); three items adapted from the scale used by Luna-Cortes and Brady (2025) to measure trust in sellers online (e.g. “I believe most online sellers are honest and trustworthy”); and three items to measure online buying intensity adapted from Forsythe et al. (2006) (e.g. “I often make purchases online”). Next, they answered the items on the CLOF scale and demographics at the end.

The results presented in Table 3 demonstrate the reliability and validity of the scales.

Next, Table 4 shows how the variance shared between each pair of constructs (squared correlation) was below the corresponding variance extracted indices. This means that the discriminant validity can be accepted.

The methodology of the structural equation modeling (SEM) was used with AMOS (SPSS) to evaluate the structural model and to estimate the set of coefficients for the relationships between variables. More precisely, a covariance structure analysis was performed, using maximum likelihood as the estimator (SEM-ML). Table 5 shows that all the hypotheses were supported.

As can be observed in Table 5, CLOF is negatively associated with the consumers' perceived financial risk of buying online, positively associated with trust in online sellers and with online buying intensity. In addition, perceived financial risk is negatively associated with consumers' online buying intensity, and trust in sellers is positively associated with consumers' online buying intensity.

This research was motivated by the need for a new scale to capture knowledge and skills that help consumers identify and be protected against online fraud. Recent studies indicated that highly educated consumers can be victims of online scams (Knuth and Ahrholdt, 2023), suggesting that higher level education by itself does not protect consumers against online fraud. In addition, existing scales did not capture the knowledge consumers require to protect themselves from online fraud. In response to this need, the CLOF multidimensional scale was developed and validated. Furthermore, the results showed that CLOF is negatively associated with perceived financial risk when buying online, and positively associated with trust in online sellers and with online buying intensity. Hence, this type of knowledge is associated with a perceived protective financial role in online contexts and influences key variables to understand participation in e-commerce. These results have both theoretical and practical implications, as explained next.

This research makes several contributions to the literature on consumer behavior, literacy, and internet research. First, the literature review revealed that existing scales (e.g. financial, digital, media, etc.) were not designed to capture the knowledge required to prevent and respond to online fraud. For instance, financial literacy is commonly measured using objective knowledge tests focused on concepts such as interest rates, inflation, and risk diversification, as in the “Big 3” framework (e.g. He et al., 2025; Pelawi et al., 2025; Rey-Ares et al., 2024; Zheng et al., 2024), while digital literacy scales predominantly focus on technical and communication skills, including dimensions such as knowledge to navigate the internet, social and emotional abilities, and the capacity to understand content (e.g. Erdem et al., 2023; Ng, 2012). Although these literacy scales are important for making financial decisions and for measuring consumers' ability to access and interact with digital information, they do not directly capture whether consumers know how to identify fraudulent online sellers or scam cues. The present research developed and validated a new multidimensional scale ready to be used in future research to capture this knowledge.

Second, this research contributes to the literature by clarifying the role of knowledge in connection with consumer perceptions of risk. Prior studies have reported inconsistent relationships between literacy and perceived risk, often finding that higher knowledge increases awareness and, consequently, perceived risk (e.g. Cheung et al., 2015; San Martín et al., 2011; Zhang and Hou, 2017). However, other studies suggest the opposite pattern (e.g. Ha and Pan, 2018; Maziriri and Chuchu, 2017). The present research offers a clearer explanation of how knowledge influences consumers' perceptions of risk, in which both the type of knowledge (e.g. online fraud) and the risk considered (e.g. financial risk when buying online) can interact with how knowledge increases or reduces perceived risk. Rather than assuming that knowledge has a uniform effect on perceived risk, our findings suggest that the relationship depends on how that knowledge functions in practice. In some cases, knowledge may simply increase awareness of potential threats, making consumers more sensitive to risks. In other cases, however, knowledge provides consumers with the tools and confidence to manage and avoid those risks. Considering this, CLOF operates as a perceived protective tool, helping consumers feel more capable of preventing financial harm when buying online. Research should consider this “awareness of risks vs. protective role of knowledge” when examining the relationship between literacy and perceived risks in future research.

Moreover, the present research advances theory by demonstrating how domain-specific literacy influences another important construct—trust. In fact, some studies suggest that higher literacy may increase skepticism toward online environments, decreasing trust (e.g. Agag and El-Masry, 2017), while others indicate that greater knowledge is positively associated with this construct (e.g. Papadopoulou et al., 2001; Pizzutti and Fernandes, 2010; Riquelme and Roman, 2014). The findings of the present research indicate that knowledge about potential fraud helps reduce uncertainty, improving consumers' ability to evaluate whether a seller presented online—an environment where there is a high risk of scams—is legitimate and trustworthy. Thus, rather than simply increasing general confidence, this type of knowledge allows consumers to make more informed judgments about whom to trust. These results help to reconcile some mixed prior findings, as the effect of knowledge on trust appears to depend on whether consumers are given the tools to assess credibility, rather than acquiring knowledge that helps to increase skepticism. In practice, online knowledge, especially the kind required to protect consumers from fraud, should focus on enabling consumers to identify risks associated with scams, rather than informing simply of potential risks. This connects with the practical implications of this research, explained next.

The CLOF scale offers a tool that can enable policymakers to avoid general assumptions about consumer vulnerability and, instead, identify specific needs for knowledge related to online fraud. By administering the scale at the population or regional level, public institutions can determine which dimensions of CLOF are underdeveloped (e.g. Knowledge of how to report fraud or use of protective tools). This can allow policymakers to design targeted public awareness campaigns that address concrete needs rather than relying on general warnings. For example, if a survey shows that citizens do not know how to report fraud properly, governments can focus on providing clearer information about the steps to report it. This may include clearer instructions, simpler reporting systems, and better visibility of official channels on public websites. However, if they find lower knowledge of how to verify the legitimacy of online companies, interventions can focus on this factor, providing key information to distinguish legitimate companies from fraudulent ones online.

From an educational perspective, the CLOF scale can be used by schools, universities, or institutions of adult education to develop fraud-prevention courses. The scale can be used to assess different groups, and each dimension can be translated into specific learning modules (e.g. verifying seller legitimacy, recognizing red flags, or understanding identity theft risks). In addition, the scale can be administered before and after educational interventions to assess learning outcomes, indicating which aspects of fraud literacy improved and which remained weak and would require further training.

For organizations involved in consumer protection—such as consumer advocacy groups, financial institutions, and e-commerce platforms—the CLOF scale can be used to segment consumers and design targeted interventions more effectively. Rather than relying only on demographic characteristics, these organizations can identify consumers at higher risk based on specific gaps in fraud-related knowledge, and adapt their recommendations accordingly. In addition, firms and platforms can also use CLOF data to improve transparency, strengthen trust signals, and design consumer warnings that help users make informed decisions. This will increase trust in sellers and institutions, which is much needed in today's society.

This research presents some limitations. First, although our findings show that CLOF is negatively associated with perceived financial risk, prior literature suggests that the relationship between this knowledge and perceived risk may not always follow this pattern. Studies have shown that more knowledgeable consumers may actually perceive higher levels of risk when buying online. This is because they possess greater awareness of potential threats (Cheung et al., 2015; San Martín et al., 2011; Zhang and Hou, 2017). Therefore, future studies should investigate whether CLOF is positively or negatively associated with perceived risk across different risk domains.

Similarly, prior literature suggests that the relationship between consumers' knowledge and trust may not always be positive and may depend on the specific trust dimension considered. For certain forms of trust during online transactions, greater knowledge may lead to increased skepticism and lower trust (Agag and El-Masry, 2017). Future research should examine whether CLOF enhances or diminishes trust across different trust targets (e.g. websites, platforms, and payment systems). This future research can reveal situations in which knowledge about potential fraud online may decrease trust rather than alleviate it, providing new practical and theoretical implications.

Although the CLOF scale demonstrates reliability and validity across different samples, and we conducted CMV tests, some items (e.g. reading terms and conditions) may be susceptible to social desirability bias. In fact, respondents might overstate some responsible behaviors. Consequently, the scale may capture perceived or self-assessed fraud literacy rather than actual performance in real fraud situations. Future research could complement the CLOF scale with scenario-based or case-based measures. Studies could ask respondents to apply their knowledge in realistic contexts, such as identifying suspicious URLs, evaluating fraudulent offers, or judging the legitimacy of online sellers based on concrete cues.

An additional limitation is associated with the context considered when developing the items of the scale. Certain items implicitly reflect regulatory and protection practices that are more salient in the US and Europe. Fraud practices and reporting systems vary across countries. Thus, some items may be interpreted differently in other regions. Future research could focus on cross-national applicability, rephrasing items using adapted language, or testing equivalence in other countries. Moreover, the data were collected at a single point in time, which limits the ability to draw causal inferences or establish the directionality of the associations. Future research should employ longitudinal or experimental designs to assess causality, and to examine how changes in CLOF over time are associated with changes in risk perceptions, trust, and online purchasing behavior.

Finally, the proposed structural model did not explicitly control for several individual differences that may affect perceived financial risk, trust in online sellers, and online buying behavior. For instance, online shopping experience, previous fraud victimization, digital competence, general risk propensity, and frequency of online purchases can interact with consumers' risk perceptions in digital environments. Although the focus of the present study was on developing and validating the literacy scale, future research should extend the model by incorporating these variables as controls or moderators. Doing so will clarify how knowledge, experience, and individual characteristics, together, are associated with online consumer behavior.

The supplementary material for this article can be found online.