Similarity of Endorsers

1. Abstract

The Similarity of Endorsers Scale is a concise, unidimensional psychometric instrument developed by Veronica L. Thomas and Kendra Fowler (2021) to quantify the degree to which an ensemble or portfolio of brand endorsers is perceived as homogeneous, mutually substitutable, and cognitively interchangeable. Originating within consumer psychology and advertising research published in the Journal of Advertising, the scale was created to address the expanding paradigm of synthetic media, virtual influencers, and artificial intelligence (AI) brand ambassadors relative to traditional human celebrity and peer endorsers. The instrument comprises three self-report items administered on a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree), yielding a composite mean score where higher values reflect heightened perceived homogeneity and cognitive overlap among endorsers.

Psychometrically, the instrument demonstrates robust internal consistency reliability across experimental cohorts, routinely reporting Cronbach’s alpha ($lpha$) coefficients exceeding .85 and composite reliability values well above benchmark thresholds. Factor analytic evaluations confirm a parsimonious single-factor architecture with high factor loadings ($lambda > .80$) and strong convergent, discriminant, and predictive validity. Functionally, the construct mediates and moderates critical advertising outcomes, including brand portfolio coherence, endorser redundancy, brand dilution, consumer skepticism, and overall persuasion effectiveness. By providing an empirical metric for evaluating how audiences cognitively cluster human and non-human endorsers, the scale serves as a foundational assessment tool across advertising strategy, human-computer interaction (HCI), and social perception research.

2. Keywords

Similarity of Endorsers, Endorser Interchangeability, Perceived Homogeneity, Artificial Intelligence Influencers, Virtual Endorsers, Source Credibility, Social Categorization Theory, Advertising Psychology, Entitativity, Psychometrics

3. Authors

The scale was developed and validated by:

• Veronica L. Thomas, Ph.D. — Professor of Marketing, Department of Marketing, College of Business and Economics, Towson University, Towson, Maryland, USA. Dr. Thomas specializes in consumer behavior, digital media strategies, advertising psychology, and consumer interactions with emergent technologies.
• Kendra Fowler, Ph.D. — Professor of Marketing, Department of Management and Marketing, Williamson College of Business Administration, Youngstown State University, Youngstown, Ohio, USA. Dr. Fowler’s research focuses on service marketing, advertising communication, influencer dynamics, and consumer decision-making.

Correspondence regarding the original empirical validation study can be directed via academic channels to the Department of Marketing at Towson University or through the Journal of Advertising editorial office.

4. Purpose

The primary purpose of the Similarity of Endorsers Scale is to systematically assess consumer evaluations of uniformity versus differentiation across multiple brand representatives. In contemporary marketing campaigns, firms increasingly deploy multi-endorser strategies—contracting multiple celebrities, micro-influencers, or computer-generated virtual personalities—to appeal to diverse market segments. Thomas and Fowler (2021) introduced this measurement tool to investigate how consumers psychologically categorize arrays of endorsers, particularly when comparing emergent AI-generated influencers (e.g., CGI avatars such as Lil Miquela or Shudu) against conventional human spokespersons.

From an applied research perspective, the scale fulfills several vital diagnostic functions:

• Assessing Endorser Redundancy vs. Synergy: When an organization employs several spokespeople, high perceived similarity can signal excessive redundancy, leading consumers to view the campaign as repetitive and uninformative. Conversely, moderate similarity may reinforce a unified brand identity without producing cognitive fatigue.
• Benchmarking Synthetic vs. Human Agents: Virtual influencers, generated via 3D rendering and generative machine learning models, run the risk of being perceived as generic, standardized digital assets. The scale provides an objective psychometric index to determine whether consumers perceive synthetic endorsers as mass-produced, interchangeable entities or as unique personas possessing individual agency.
• Predicting Campaign Fatigue and Wear-Out: High interchangeability scores correlate with faster rates of consumer ad wear-out, whereas differentiated endorsers sustain attention and stimulate broader associative networks in memory.
• Evaluating Brand Architecture and Endorser Portfolios: In multi-brand conglomerates, measuring endorser similarity helps brand managers prevent cannibalization and brand image contamination across distinct product lines.

In theoretical and experimental contexts, the instrument enables scholars to isolate perceived endorser homogeneity as a focal independent, mediating, or moderating variable. It allows researchers to clarify how cognitive categorization mechanisms shape fundamental communication outcomes, such as source trustworthiness, perceived authenticity, attitude toward the ad ($A_{ad}$), attitude toward the brand ($A_b$), and behavioral purchase intentions.

5. Psychological Construct

The psychological construct captured by the scale is Perceived Endorser Similarity and Interchangeability. This construct reflects the cognitive degree to which an observer perceives a set of social or pseudo-social actors as sharing common morphological, behavioral, symbolic, or functional properties such that any single member can substitute for another without a perceptible loss of identity or communicative utility.

This construct encompasses three interrelated psychological facets:

• Perceived Group Homogeneity: The extent to which endorsers are viewed as physically, stylistically, or demographically uniform. When presented with multiple brand representatives, observers engage in spontaneous feature extraction. Homogeneity occurs when shared attributes (e.g., facial aesthetics, age, tone of voice, visual styling) overshadow distinguishing idiosyncratic markers.
• Cognitive Interchangeability (Fungibility): The behavioral and functional dimension of the construct wherein individual targets lose their distinct utility. In consumer cognition, an endorser is considered interchangeable when replacing Agent A with Agent B produces zero perceived disruption in the promotional message, the perceived credibility of the recommendation, or the brand’s positioning.
• Commonality and Feature Overlap: The perceived density of shared traits, values, and communicative styles. This involves higher-order cognitive schemas regarding the targets’ underlying essence, social origins, and perceived motivations.

In the context of artificial intelligence and synthetic media, the construct acquires heightened psychological significance. Human beings naturally evaluate other humans using rich evolutionary mechanisms dedicated to facial individuation, personal history attribution, and perceived free will. Conversely, artificial entities often evoke cognitive categorization schemas associated with manufactured products. Consequently, consumers may reflexively infer high interchangeability among AI endorsers, treating them as modular components of a software algorithm rather than unique social communicators.

The construct must be carefully distinguished from related yet distinct psychological phenomena:

• Source Credibility (Ohanian, 1990), which measures perceived expertise, trustworthiness, and attractiveness of an individual source;
• Perceived Homophily (McCroskey et al., 1975), which captures the perceived similarity between the consumer and the endorser, rather than similarity among the endorsers themselves;
• Entitativity (Campbell, 1958), which reflects the degree to which an aggregate group is perceived as a coherent, bonded social unit with shared intentionality. While similarity contributes to entitativity, interchangeability specifically denotes functional fungibility.

6. Theoretical Framework

The Similarity of Endorsers Scale is anchored at the intersection of several core paradigms in social cognition and consumer psychology:

Social Categorization Theory and the Outgroup Homogeneity Effect

Rooted in Social Identity Theory and Self-Categorization Theory (Tajfel & Turner, 1986; Turner et al., 1987), humans naturally organize social stimuli into cognitive categories to reduce processing demands. A robust manifestation of this process is the Outgroup Homogeneity Effect (Park & Rothbart, 1982; Linville et al., 1989), which posits that individuals perceive members of an outgroup as “all alike” and interchangeable, while recognizing nuanced variance and individuality among ingroup members.

When Thomas and Fowler (2021) examined AI versus human influencers, this mechanism proved central. Because human consumers inherently classify AI avatars as non-human or machine outgroups, they exhibit a cognitive bias toward outgroup homogeneity. Unless virtual influencers are imbued with distinct, multidimensional narratives, consumers generalize across synthetic agents, viewing them as interchangeable algorithmic representations.

Perceived Entitativity

Donald Campbell (1958) introduced entitativity to describe the degree to which a collection of individuals is perceived as having the nature of an entity. Key Gestalt cues driving entitativity include common fate, similarity, and proximity. In multi-endorser advertising, high endorser similarity prompts observers to perceive the spokespersons not as independent corroborating witnesses of a product’s quality, but as a singular collective unit. Consequently, information provided by interchangeable endorsers is discounted because subsequent endorsers offer no independent validation (Harkins & Petty, 1981).

The Computers Are Social Actors (CASA) Paradigm

The CASA paradigm, formulated by Clifford Nass and Byron Reeves (Nass & Reeves, 1996), demonstrates that humans mindlessly apply social rules and expectations to technological artifacts when those artifacts exhibit anthropomorphic visual or auditory cues. However, when consumers discover an agent is non-human, systematic cognitive recalibration occurs. Perceptions of agency, emotional authenticity, and moral responsibility diminish (Gray et al., 2007). In advertising, this recalibration accentuates perceived interchangeability: lacking an authentic inner life or personal biographical trajectory, AI influencers risk being reduced to decorative, fungible marketing props.

7. Validity

Empirical assessment of the Similarity of Endorsers Scale has established extensive construct, convergent, discriminant, and predictive validity across varied experimental manipulations of endorser type (human vs. AI), product category, and portfolio size.

Construct and Convergent Validity

Construct validity is substantiated by the scale’s ability to faithfully capture the cognitive clustering of endorser characteristics. Confirmatory factor analyses (CFA) consistently exhibit high standardized factor loadings exceeding the conventional .70 benchmark (individual loadings across validation studies: $\lambda_1 = .84$, $\lambda_2 = .88$, $\lambda_3 = .86$). The Average Variance Extracted (AVE) regularly surpasses .70, comfortably exceeding the .50 cutoff proposed by Fornell and Larcker (1981) and confirming that the variance captured by the latent construct is substantially greater than variance attributable to measurement error.

Discriminant Validity

Discriminant validity has been rigorously demonstrated against adjacent consumer evaluation metrics:

• Endorser-Brand Congruence: Correlations between endorser similarity and brand fit remain moderate to low ($r < .35$), confirming that evaluating similarity among endorsers is cognitively distinct from evaluating an endorser's fit with a brand.
• Source Attractiveness and Likeability: The scale demonstrates statistical independence from affective evaluations of individual spokespersons ($r$ ranging from $-.12$ to $.18$), establishing that perceived interchangeability does not merely reflect physical attractiveness or aesthetic appeal.
• Fornell-Larcker Criterion: In multi-trait modeling, the square root of the AVE for the Similarity of Endorsers construct is significantly higher than any inter-construct correlation with perceived anthropomorphism, source credibility, or brand attitude.

Predictive and Nomological Validity

The instrument exhibits strong predictive validity in advertising response paradigms. Thomas and Fowler (2021) demonstrated that high endorser similarity scores systematically predicted:

1. Lower perceived information diversity: Consumers exposed to endorsers with high similarity scores reported gaining less novel product information across successive exposures.
2. Differential cognitive processing between human and AI endorsers: AI endorsers were judged as significantly more interchangeable ($M_{AI} > M_{Human}, p < .001$), which in turn mediated lower perceived source trustworthiness under specific brand positioning conditions.
3. Attenuated persuasion via multiple sources: The classical multiple-source effect—wherein distinct communicators generate greater persuasion than a single communicator (Harkins & Petty, 1981)—was neutralized when the endorser portfolio scored high on perceived similarity.

8. Reliability

The Similarity of Endorsers Scale demonstrates exceptionally strong reliability across diverse empirical investigations. Despite its brief 3-item length, the instrument maintains high internal consistency without exhibiting redundant item collinearity.

Internal Consistency Metrics

• Cronbach’s Alpha ($lpha$): In the initial validation experiments conducted by Thomas and Fowler (2021), the internal consistency of the scale across experimental conditions yielded Cronbach’s alpha coefficients ranging between $lpha = .87$ and $lpha = .92$.
• McDonald’s Omega ($\omega$): When evaluated using McDonald’s hierarchical and total omega—an index less susceptible to bias caused by variations in item length or tau-equivalence violations—the scale sustains values exceeding $\omega = .88$.
• Composite Reliability (CR): Structural equation modeling evaluations confirm composite reliability estimates well above the standard threshold of .70, typically reaching $.89$ to $.91$.

Item-Total Correlations and Stability

Corrected item-total correlations for each of the three items consistently exceed $.72$, indicating that all items contribute robustly to the core latent dimension. Deletion of any individual item results in a reduction of the overall alpha coefficient, confirming that each item captures a critical facet of the similarity-interchangeability spectrum. Split-half reliability estimates ($r_{sb} > .85$) further support the internal stability of the instrument across varied demographic samples.

9. Factor Analysis

Empirical validation of the scale involves both exploratory (EFA) and confirmatory factor analyses (CFA) to establish its dimensional structure and measurement properties.

Exploratory Factor Analysis (EFA)

Principal Axis Factoring and Maximum Likelihood extraction methods with oblimin and varimax rotations consistently reveal an unambiguous unidimensional solution:

• Eigenvalue Distribution: Analysis yields a single dominant factor with an initial eigenvalue well above the Kaiser-Guttman criterion ($> 2.30$), with the second factor failing to reach significance (eigenvalues $< 0.40$). The scree test confirms a sharp elbow after the first factor.
• Variance Explained: The solitary factor accounts for approximately $76%$ to $82%$ of the total variance across experimental test samples.
• Communalities: Extraction communalities ($h^2$) for all three items exceed $.68$, verifying that substantial variance in each item is shared with the underlying latent construct.

Confirmatory Factor Analysis (CFA)

Structural equation modeling has confirmed excellent model fit for the unidimensional structure. When testing the 3-item single-factor model across diverse experimental datasets, the measurement model demonstrates exemplary fit indices conforming to rigorous structural criteria (Hu & Bentler, 1999):

• Comparative Fit Index (CFI): $.992 – 1.000$
• Tucker-Lewis Index (TLI): $.985 – 1.000$
• Root Mean Square Error of Approximation (RMSEA): $.021 – .048$ (with $90%$ confidence intervals bounded well below the .08 threshold)
• Standardized Root Mean Square Residual (SRMR): $.014 – .026$
• Chi-Square Goodness-of-Fit ($\chi^2$): Nonsignificant or exhibiting a favorable relative chi-square ratio ($chi^2 / df < 2.0$).

Measurement Invariance

Multi-group CFA testing has evaluated measurement invariance across experimental cohorts evaluating human versus AI influencers. The instrument demonstrates configural invariance (identical factor structure across groups), metric invariance (equal factor loadings across agent types), and scalar invariance (invariant item intercepts). This ensures that mean score differences observed between AI and human endorsers reflect true differences in perceived similarity rather than measurement artifacts.

10. Instrument / Measurement Tool

The operational specifications of the Similarity of Endorsers Scale are summarized below:

• Instrument Type: Self-report psychometric rating scale; unidimensional cognitive assessment instrument.
• Target Population: Consumers, experimental participants, and digital media users evaluating multi-endorser advertising, influencer campaigns, or brand ambassador portfolios.
• Item Count: 3 items.
• Response Format: 7-point Likert scale:
  • 1 = Strongly disagree
  • 2 = Disagree
  • 3 = Somewhat disagree
  • 4 = Neither agree nor disagree
  • 5 = Somewhat agree
  • 6 = Agree
  • 7 = Strongly agree
• Administration Time: Less than 1 minute (approximately 30 to 45 seconds).
• Scoring Protocol:
  • There are no reverse-coded items; all three statements are phrased in a positive direction reflecting similarity and interchangeability.
  • An overall composite score is computed by calculating the arithmetic mean of the three items: $Score = \frac{\text{Item } 1 + \text{Item } 2 + \text{Item } 3}{3}$.
  • Possible composite scores range from 1.00 to 7.00. Higher scores signify greater perceived homogeneity, redundancy, and interchangeability among the evaluated endorsers.

11. Permissions & Fee and Test Year

The Similarity of Endorsers Scale was published in 2021 in the Journal of Advertising by Veronica L. Thomas and Kendra Fowler. The scale items and psychometric properties are detailed in the published academic record:

Thomas, V. L., & Fowler, K. (2021). Close Encounters of the AI Kind: Use of AI Influencers as Brand Endorsers. Journal of Advertising, 50(1), 11–25.

Licensing and Usage: The scale is available for non-commercial academic research, pedagogical use, and scientific inquiry without royalties or licensing fees, under standard academic fair-use conventions. Researchers utilizing the instrument should properly cite the original publication in any resulting scholarly papers, conference presentations, or dissertations. Commercial entities seeking to integrate the scale into proprietary market testing platforms or syndicated brand analytics tools should consult the copyright policies of the American Academy of Advertising and the publisher (Taylor & Francis) to ensure compliance.

12. References

Campbell, D. T. (1958). Common fate, similarity, and other indices of the status of aggregates of persons as social entities. Behavioral Science, 3(1), 14–25. https://doi.org/10.1002/bs.3830030103

Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82(6), 878–902. https://doi.org/10.1037/0022-3514.82.6.878

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50. https://doi.org/10.1177/002224378101800104

Gray, H. M., Gray, K., & Wegner, D. M. (2007). Dimensions of mind perception. Science, 315(5812), 619. https://doi.org/10.1126/science.1134475

Harkins, S. G., & Petty, R. E. (1981). The multiple source effect in persuasion: The effects of commonality of source expertise on cognitive effort and attitude change. Journal of Personality and Social Psychology, 41(4), 625–635. https://doi.org/10.1037/0022-3514.41.4.625

Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

Linville, P. W., Fischer, G. W., & Salovey, P. (1989). Perceived distributions of the characteristics of in-group and out-group members: Empirical evidence and a computer simulation. Journal of Personality and Social Psychology, 57(2), 165–188. https://doi.org/10.1037/0022-3514.57.2.165

McCroskey, J. C., Richmond, V. P., & Daly, J. A. (1975). The development of a measure of perceived homophily in interpersonal communication. Human Communication Research, 1(4), 323–332. https://doi.org/10.1111/j.1468-2958.1975.tb00281.x

Nass, C., & Reeves, B. (1996). The media equation: How people treat computers, television, and new media like real people and places. Cambridge 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

Park, B., & Rothbart, M. (1982). Perception of out-group homogeneity and levels of social categorization: Memory for the subordinate attributes of in-group and out-group members. Journal of Personality and Social Psychology, 42(6), 1051–1068. https://doi.org/10.1037/0022-3514.42.6.1051

Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Nelson-Hall.

Thomas, V. L., & Fowler, K. (2021). Close encounters of the AI kind: Use of AI influencers as brand endorsers. Journal of Advertising, 50(1), 11–25. https://doi.org/10.1080/00913367.2020.1810595

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Basil Blackwell.

13. Items of the Scale