eWOM Credibility Scale (eWOM-CRED)

1. Abstract

The eWOM Credibility Scale (eWOM-CRED) is a multidimensional psychometric instrument formulated to assess consumer perceptions of the veracity, dependability, and informational value of electronic word-of-mouth (eWOM) within digital information ecologies. Systematized originally in consumer informatics and communication psychology by Christy M. K. Cheung, Matthew K. O. Lee, and Neil Rabjohn (2008), the instrument synthesizes tenets of the Elaboration Likelihood Model (ELM) and dual-process information processing paradigms. It quantifies eWOM credibility across three core primary dimensions: Argument Strength (evaluating the informational quality, relevance, comprehensiveness, and persuasiveness of the consumer review text), Source Credibility (appraising the perceived expertise, trustworthiness, and objectivity of the online communicator), and Confirmation of Prior Beliefs (gauging cognitive congruence between incoming diagnostic review content and the recipient’s antecedent expectations or domain knowledge).

The operationalized measurement battery typically comprises 12 to 16 self-report items evaluated on a 7-point Likert scale anchored from 1 (“Strongly Disagree”) to 7 (“Strongly Agree”), alongside focal criterion items evaluating overarching perceived review credibility and review adoption intention. Across empirical validation studies in e-commerce, consumer review platforms, online discussion boards, and social commerce architectures, the scale demonstrates rigorous psychometric adequacy. Internal consistency reliabilities across subscales systematically surpass established benchmarks, with Cronbach’s alpha (α) and composite reliability coefficients generally ranging between 0.81 and 0.93. Confirmatory factor analyses consistently substantiate the tripartite latent structure, evidencing convergent and discriminant validity via average variance extracted indices exceeding 0.50. The scale serves as a foundational assessment tool across behavioral economics, human-computer interaction, online consumer psychology, and market surveillance research.

2. Keywords

electronic word-of-mouth, eWOM credibility, argument strength, source credibility, confirmation of prior beliefs, information adoption model, elaboration likelihood model, psychometrics, consumer reviews, online trust, epistemic vigilance

3. Authors

The structural operationalization and primary psychometric validation of the electronic word-of-mouth credibility and adoption architecture was established by:

• Christy M. K. Cheung, Ph.D. — Professor of Information Systems and e-Business Management, Department of Management, Marketing and Information Systems, Hong Kong Baptist University, Kowloon Tong, Hong Kong. Dr. Cheung is a leading international authority on consumer behavior in virtual communities, social media psychology, and online knowledge exchange systems.
• Matthew K. O. Lee, Ph.D. — Chair Professor of Information Systems and e-Commerce, Department of Information Systems, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong. Professor Lee has published widely on organizational technology adoption, electronic trust dynamics, and virtual commerce infrastructure.
• Neil Rabjohn, M.Sc. — Department of Information Systems, City University of Hong Kong. Mr. Rabjohn contributed substantially to the experimental design, quantitative modeling, and field survey instrumentation capturing online opinion adoption mechanisms.

4. Purpose

In contemporary computer-mediated environments, consumers face an unprecedented deluge of peer-generated information, characterized by asymmetric information quality, varying author anonymity, and systemic vulnerability to commercially biased, fraudulent, or automated promotional content. Unlike traditional, face-to-face word-of-mouth (WOM), where relational ties and interpersonal social cues provide intuitive heuristics for assessing veracity, electronic word-of-mouth is fundamentally decentralized, asynchronous, and socially disembodied. Consequently, consumers must engage in specialized epistemic assessment strategies to distinguish authentic user evaluations from deceptive, inaccurate, or superficial rhetoric.

The principal objective of the eWOM Credibility Scale (eWOM-CRED) is to establish a standardized, statistically validated mechanism for quantifying how individuals evaluate the epistemic integrity and cognitive trustworthiness of online peer opinions. Rather than treating review credibility as a monolithic or unidimensional construct, the instrument deconstructs the evaluation process into discrete informational, communicator-oriented, and cognitive-congruence vectors. By doing so, it serves several essential theoretical and applied functions across computational social science, commercial analytics, and regulatory policy.

From an applied research perspective, the instrument facilitates diagnostic evaluations of platform architecture. Digital marketplace designers use the scale to determine which metadata features—such as verified buyer badges, detailed reviewer profiles, historical contribution tallies, or community up-voting systems—meaningfully enhance perceived reviewer credibility versus those that induce skepticism. In market intelligence and consumer psychology, the scale illuminates the specific thresholds where negative or contradictory user reviews retain credibility, guiding risk mitigation strategies for brand reputation management. Furthermore, public interest organizations, legal researchers, and digital regulatory bodies deploy the eWOM-CRED framework to gauge the deceptive efficacy of astroturfing campaigns, undisclosed influencer endorsements, and artificial intelligence-generated consumer testimonials.

From a theoretical standpoint, the scale addresses the persistent question of cognitive information integration within virtual communities. By measuring the relative contribution of central-route properties (e.g., argumentative coherence, logical rigor) versus peripheral cues (e.g., reviewer reputation indicators) and top-down cognitive matching (prior belief congruence), the scale enables empirical testing of competing models of online attitude formation, purchase intention, and cognitive dissonance reduction.

5. Psychological Construct

The latent construct of eWOM Credibility represents an individual’s subjective appraisal of whether an online recommendation or consumer review reflects factual truth, objective reality, and authentic personal experience. Drawing from social judgment theory, cognitive consistency literature, and psycholinguistic frameworks, the scale breaks this evaluation into three interrelated yet conceptually autonomous sub-dimensions:

1. Argument Strength

Argument strength pertains to the subjective perception of the quality, validity, logical soundness, and persuasiveness of the information embedded within the message text. In digital product reviews, argument strength is operationalized through criteria such as:

• Informativeness: The review provides substantive, actionable details regarding product performance, ergonomics, failure rates, or utility beyond superficial assertions.
• Completeness: The commentary explores both positive and negative attributes, demonstrating an analytical perspective rather than a one-sided polemic.
• Clarity and Coherence: The review maintains a clear semantic structure, articulates concrete scenarios of usage, and presents evidentiary claims that rationally support the reviewer’s concluding recommendation.

Individuals evaluating a review exhibiting high argument strength process the message via central cognitive pathways, scrutinizing the validity of claims against contextual benchmarks.

2. Source Credibility

Source credibility encapsulates the recipient’s evaluation of the communicator’s characteristics, specifically their perceived competence and moral integrity. Because online reviewers are frequently anonymous or pseudo-anonymous, source credibility assessments depend heavily on system-provided metadata, behavioral badges, linguistic tone, and projected identity markers. This dimension is bifurcated theoretically into:

• Perceived Expertise: The degree to which the author is perceived to possess specialized product knowledge, technical competence, domain experience, and qualified insight necessary to formulate authoritative judgments.
• Perceived Trustworthiness: The perception that the author is impartial, objective, unmotivated by commercial incentives (e.g., sponsored postings or affiliated kickbacks), and genuinely interested in informing fellow consumers.

When source credibility is high, consumers perceive lower information asymmetry and lower transactional risk, allowing the review to serve as a reliable cognitive heuristic.

3. Confirmation of Prior Beliefs

The third dimension is rooted in cognitive psychology, specifically theories of confirmation bias and selective cognitive assimilation. Confirmation of prior beliefs captures the degree to which the diagnostic claims, affective valence, and technical evaluations presented in an eWOM communication align with the recipient’s pre-existing schemas, prior brand impressions, or experiential domain knowledge. The underlying psychological mechanisms include:

• Cognitive Consistency: Information that mirrors existing mental models generates cognitive fluency and reduced psychological friction, predisposing the reader to classify the review as accurate and rational.
• Disconfirmation Skepticism: When a consumer encounters a review that sharply departs from their established expectations—such as an overwhelmingly negative critique of a celebrated brand, or an uncritical eulogy of a notoriously defective item—the individual experiences cognitive dissonance, prompting hyper-critical evaluation or defensive rejection of the review’s credibility.

Together, these three dimensions constitute a triadic matrix: the message content (argument strength), the messenger’s perceived qualities (source credibility), and the recipient’s internal cognitive baseline (prior belief confirmation).

6. Theoretical Framework

The theoretical architecture of the eWOM Credibility Scale synthesizes two foundational paradigms in social and communication psychology: the Information Adoption Model (IAM), formulated by Sussman and Siegal (2003), and the Elaboration Likelihood Model (ELM), established by Petty and Cacioppo (1986), complemented by Carl Hovland’s classical source-credibility paradigm and Leon Festinger’s theory of cognitive dissonance.

Dual-Process Information Processing (ELM and IAM)

The Elaboration Likelihood Model posits that persuasive communication is processed via two distinct cognitive routes depending on an individual’s motivation and cognitive ability to elaborate on the message content:

• Central Route: Characterized by high cognitive elaboration, extensive scrutiny of logical arguments, critical assessment of evidence, and systematic evaluation of causal claims. In the eWOM-CRED scale, this route is operationalized through Argument Strength. Highly motivated consumers seeking to minimize monetary or functional risk engage in deep textual scrutiny of review quality.
• Peripheral Route: Characterized by low cognitive elaboration, where individuals rely on environmental, contextual, or heuristic cues to formulate rapid assessments without expending heavy cognitive resources. In the eWOM context, this route is operationalized through Source Credibility (e.g., star ratings, reviewer tenure, verified purchaser status, profile iconography).

Sussman and Siegal (2003) integrated ELM with the Technology Acceptance Model (TAM) to yield the Information Adoption Model, arguing that knowledge transfer and information adoption in virtual spaces depend primarily on informational usefulness, which is directly predicted by the combination of central (argument quality) and peripheral (source credibility) vectors. Cheung, Lee, and Rabjohn (2008) expanded this paradigm to non-organizational, customer-to-customer discussion forums by introducing the subjective internal state of the recipient, conceptualized as prior belief confirmation.

Hovland’s Source-Characteristics Paradigm

The operationalization of source credibility in eWOM-CRED draws directly from the classical Yale Communication Research Program headed by Carl Hovland and colleagues (1953). Hovland established that the effectiveness of persuasive communication depends critically on two dimensions of the communicator: perceived expertness (the communicator’s perceived capacity to make correct assertions) and trustworthiness (the degree of confidence in the communicator’s intent to communicate assertions without bias). In anonymous digital forums, consumers extract these cues from linguistic style, self-disclosed background information, and community reputation rankings.

Cognitive Dissonance and Consistency Theories

Festinger’s (1957) theory of cognitive dissonance and Sherif and Hovland’s (1961) social judgment theory provide the theoretical grounding for the third subscale. When incoming digital communication falls within an individual’s “latitude of acceptance” (i.e., confirms prior beliefs), it is assimilated smoothly, and the source is credited with good judgment. When it falls within the “latitude of rejection” (sharply disconfirming prior beliefs), consumers perceive the communication as anomalous, attributing the divergence to reviewer incompetence, malice, or idiosyncratic bias. By incorporating this construct, the eWOM-CRED scale moves beyond stimulus-only properties to incorporate the cognitive-affective state of the receiver.

7. Validity

The eWOM Credibility Scale and its derivative adaptations have undergone comprehensive psychometric evaluation, demonstrating robust construct, convergent, discriminant, and predictive validity across diverse online contexts, including electronics consumer boards, travel review sites (e.g., TripAdvisor), social commerce environments, and healthcare patient forums.

Construct and Convergent Validity

Construct validity has been established through both exploratory and confirmatory factor modeling. Cheung, Lee, and Rabjohn (2008) conducted rigorous testing of convergent validity using structural equation modeling (SEM). Convergent validity is confirmed when each measurement indicator loads significantly and substantially on its designated latent factor, with standardized factor loadings exceeding the conventional 0.70 benchmark, and when the Average Variance Extracted (AVE) for each latent dimension surpasses 0.50.

In empirical assessments:

• Standardized loadings for Argument Strength items systematically fall between 0.74 and 0.88, demonstrating that the observed variables share over 50% to 75% of their variance with the latent construct.
• Standardized loadings for Source Credibility items range from 0.72 to 0.89 across both student and broad consumer populations.
• Loadings for Confirmation of Prior Beliefs range from 0.76 to 0.86.
• Calculated AVE values across all dimensions exceed 0.58, confirming that latent construct variance substantially outweighs measurement error variance.

Discriminant Validity

Discriminant validity has been confirmed using the Fornell-Larcker criterion and the Heterotrait-Monotrait ratio of correlations (HTMT):

• Fornell-Larcker Criterion: In primary validation datasets, the square root of the AVE for Argument Strength (approx. 0.81), Source Credibility (approx. 0.83), and Confirmation of Prior Beliefs (approx. 0.82) consistently exceeds the inter-construct correlation coefficients between any pair of latent constructs (which typically range from r = 0.32 to r = 0.58).
• HTMT Ratios: In modern methodological replications, all HTMT values remain well below the conservative 0.85 threshold, substantiating that the three dimensions measure distinct psychological processes rather than overlapping variance artifacts.

Nomological and Predictive Validity

Nomological and criterion-related predictive validity is evidenced by the scale’s capacity to account for variance in downstream behavioral intentions. Structural equation models demonstrate that Argument Strength and Source Credibility exert robust, statistically significant direct effects on overall Perceived eWOM Credibility (path coefficients typically: β = 0.34 to 0.48, p < 0.001 for Argument Strength; β = 0.22 to 0.38, p < 0.001 for Source Credibility). Perceived eWOM Credibility, in turn, acts as a primary mediator driving Information Adoption (β = 0.45 to 0.62, p < 0.001) and subsequent Purchase Intention (accounting for upwards of 40% to 55% of total variance explained, R² = 0.42–0.56).

8. Reliability

The reliability of the eWOM Credibility Scale has been confirmed via multiple statistical metrics across diverse international cohorts and technological platforms.

Internal Consistency

Internal consistency is typically evaluated using both Cronbach’s alpha (α) and Composite Reliability (CR). Empirical studies employing the Cheung et al. (2008) operationalization routinely report values well in excess of the recommended 0.70 threshold:

• Argument Strength Subscale: Cronbach’s α values range between 0.84 and 0.91 across independent studies; Composite Reliability ranges between 0.86 and 0.92.
• Source Credibility Subscale: Cronbach’s α values range between 0.82 and 0.89; Composite Reliability ranges between 0.85 and 0.90.
• Confirmation of Prior Beliefs Subscale: Cronbach’s α values range between 0.80 and 0.88; Composite Reliability ranges between 0.83 and 0.89.
• Overall Perceived eWOM Credibility (Criterion Scale): Cronbach’s α values range between 0.88 and 0.93; Composite Reliability ranges between 0.90 and 0.94.

Item-Total Correlations and Stability

Corrected item-total correlations across all operationalized items consistently surpass 0.55, indicating strong item-level discrimination and high internal coherence without redundancy. In experimental laboratory settings where test-retest reliability across a two-week interval was examined using static product review stimuli, the intraclass correlation coefficients (ICC) exceeded 0.78, demonstrating stability of the instrument when stimulus properties remain constant.

9. Factor Analysis

The dimensional topology of the eWOM Credibility Scale has been extensively evaluated using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

Exploratory Factor Analysis (EFA)

During initial instrument validation, EFA utilizing Principal Axis Factoring (PAF) or Principal Component Analysis (PCA) with Promax or Varimax rotation confirmed a clear three-factor structure underlying the antecedent dimensions, alongside a separate criterion factor for perceived eWOM credibility:

• Kaiser-Meyer-Olkin (KMO) measures of sampling adequacy consistently exceed 0.85 (typically KMO = 0.88–0.92), verifying sample suitability for factor extraction.
• Bartlett’s Test of Sphericity yields highly significant outcomes (p < 0.0001), indicating adequate correlation matrices.
• Extracted factors account for over 65% to 74% of the total cumulative variance.
• Items exhibit primary factor loadings above 0.70, with cross-loadings across alternative factors remaining well below the 0.30 threshold.

Confirmatory Factor Analysis (CFA)

Subsequent testing via maximum-likelihood Confirmatory Factor Analysis rigorously substantiated the hypothesized three-factor oblique model over alternative competing models (such as single-factor models or two-factor models collapsing argument strength and source credibility). Fit indices reported in foundational and replication studies systematically satisfy standard psychometric cutoffs:

• Normed Chi-Square (χ²/df): Typically ranges from 1.35 to 2.42 (benchmark: < 3.0).
• Comparative Fit Index (CFI): Ranges between 0.95 and 0.98 (benchmark: ≥ 0.95).
• Tucker-Lewis Index (TLI / NNFI): Ranges between 0.94 and 0.97 (benchmark: ≥ 0.95).
• Root Mean Square Error of Approximation (RMSEA): Ranges from 0.038 to 0.058, with 90% confidence intervals bounded below 0.07 (benchmark: < 0.06).
• Standardized Root Mean Square Residual (SRMR): Ranges from 0.031 to 0.049 (benchmark: < 0.08).

Comparison with a one-factor common method bias model (Harman’s single-factor CFA) indicates significantly poorer fit (e.g., χ²/df > 8.50, CFI < 0.65, RMSEA > 0.14), rejecting the hypothesis of monolithic unmeasured artifactual variance.

10. Instrument / Measurement Tool

The structural characteristics, administration protocols, and scoring modalities of the eWOM Credibility Scale are detailed below:

• Instrument Name: Electronic Word-of-Mouth Credibility Scale (eWOM-CRED / IAM Review Credibility Battery).
• Test Type: Multi-item psychometric self-report questionnaire designed for online surveys, field assessments, and laboratory experiments.
• Target Population: Consumers, internet users, and digital forum participants aged 18 years and above evaluating online product or service reviews.
• Primary Dimensions / Subscales:
  • Subscale 1: Argument Strength (AS): Evaluates perceived informational quality, persuasiveness, and depth of the review (typically 4 items).
  • Subscale 2: Source Credibility (SC): Evaluates perceived expertise, competence, and trustworthiness of the reviewer (typically 4 items).
  • Subscale 3: Confirmation of Prior Beliefs (CPB): Evaluates consistency with antecedent knowledge and subjective expectations (typically 3–4 items).
  • Criterion Scale: Overall eWOM Credibility (CRED): Measures direct perceived truthfulness, believability, and factual accuracy of the review (typically 4 items).
• Total Item Count: 15 items across the four sub-batteries (standard comprehensive format).
• Response Format: 7-point Likert scale:
  • 1 = Strongly Disagree
  • 2 = Disagree
  • 3 = Somewhat Disagree
  • 4 = Neutral (Neither Agree nor Disagree)
  • 5 = Somewhat Agree
  • 6 = Agree
  • 7 = Strongly Agree
• Scoring and Computational Rules:
  • All items are framed in a positive semantic direction; reverse scoring is not required in the standard edition.
  • Subscale scores are computed by calculating the arithmetic mean of the respective subscale items (Sum of item ratings divided by number of items). This preserves the intuitive 1–7 scale interpretation.
  • Scores from 1.00 to 2.99 indicate low perceived standing (e.g., weak arguments, low source credibility, or strong disconfirmation).
  • Scores from 3.00 to 4.99 indicate moderate or equivocal perceptions.
  • Scores from 5.00 to 7.00 reflect high argument strength, authoritative source credibility, strong cognitive confirmation, or robust overall credibility.
  • For structural equation modeling and regression path analyses, subscale scores are treated as continuous latent variables or composite observed indicators predicting downstream adoption.
• Administration Time: Approximately 3 to 6 minutes for completion when paired with an assigned digital review stimulus.

11. Permissions & Fee and Test Year

The foundational empirical architecture and measurement items were first published in 2008 by Christy M. K. Cheung, Matthew K. O. Lee, and Neil Rabjohn in the peer-reviewed scholarly journal Internet Research (Emerald Group Publishing). As an academic psychometric instrument adapted from theoretical constructs in information systems and cognitive psychology, the scale is generally accessible without financial cost for non-commercial academic research, pedagogical purposes, and university-based scientific studies, provided appropriate academic attribution and citation are rendered.

Commercial utilization, deployment within proprietary market research panels, or integration into commercial reputation-monitoring software may require direct inquiry and permission from the copyright holder (Emerald Group Publishing) or the original scale authors. Researchers intending to adapt the scale for non-English linguistic populations should conduct rigorous forward-backward translation and cross-cultural structural validation before deployment.

12. References

The following foundational peer-reviewed studies document the theoretical inception, psychometric validation, and empirical application of the eWOM Credibility Scale:

• Bhattacherjee, A., & Sanford, C. (2006). Influence processes for information technology acceptance: An elaboration likelihood model. MIS Quarterly, 30(4), 805–825. https://doi.org/10.2307/25148755
• Cheung, C. M. K., Lee, M. K. O., & Rabjohn, N. (2008). The impact of electronic word-of-mouth: The adoption of online opinions in online customer communities. Internet Research, 18(3), 229–247. https://doi.org/10.1108/10662240810883290
• Cheung, C. M. K., & Thadani, D. R. (2012). The impact of electronic word-of-mouth communication: A literature analysis and integrative model. Decision Support Systems, 54(1), 461–470. https://doi.org/10.1016/j.dss.2012.06.008
• Festinger, L. (1957). A theory of cognitive dissonance. Stanford University Press. https://doi.org/10.1515/9781503620766
• Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and persuasion: Psychological studies of opinion change. Yale University Press.
• Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers’ perceived expertise, trustworthiness, and attractiveness. Journal of Advertising, 19(3), 39–52. https://doi.org/10.1080/00913367.1990.10673191
• Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In Advances in Experimental Social Psychology (Vol. 19, pp. 123–205). Academic Press. https://doi.org/10.1016/S0065-2601(08)60214-2
• Prendergast, G., Ko, D., & Yuen, S. Y. V. (2010). Online word of mouth and consumer purchase intentions. International Journal of Advertising, 29(5), 687–708. https://doi.org/10.2501/S0265048710201427
• Sherif, M., & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in communication and attitude change. Yale University Press.
• Sussman, S. W., & Siegal, W. S. (2003). Informational influence in organizations: An integrated approach to knowledge adoption. Information Systems Research, 14(1), 47–65. https://doi.org/10.1287/isre.14.1.47.14767

13. Items of the Scale