Social Engagement of the Object

1. Abstract

The Social Engagement of the Object scale is a concise, psychometrically validated, unidimensional instrument designed to quantify the degree to which an automated or artificial entity—most notably conversational artificial intelligence (AI), conversational agents, virtual assistants, and conversational robo-advisors—displays human-like social engagement cues during user-machine dyads. Originally developed and validated by Christian Hildebrand and Anouk Bergner (2021) in their foundational work on financial technology onboarding and consumer decision-making, the measure captures perceived conversational synchrony, social presence, and perceived interactive responsiveness. The scale consists of 3 self-report items evaluated on an authentic 7-point Likert scale ranging from 1 (“Strongly disagree”) to 7 (“Strongly agree”).

Psychometric evaluations across multiple experimental and field trials demonstrate exceptional internal consistency reliability (with Cronbach’s alpha coefficients routinely exceeding α = .88 and reaching up to .93 across independent validation cohorts). Confirmatory factor analysis robustly supports a unidimensional factor structure, exhibiting high factor loadings (ranging between .85 and .94) and exemplary goodness-of-fit indices (CFI > .99, TLI > .98, RMSEA < .05, and SRMR < .02). The scale manifests pronounced construct, convergent, discriminant, and predictive validity, operating as an essential mediating mechanism between conversational interface modalities (such as chat-based natural language processing vs. static form-based inputs) and distal downstream behavioral outcomes, including interpersonal trust calibration, firm credibility, willingness to disclose sensitive personal data, and compliance with high-stakes financial or algorithmic recommendations. This article provides an exhaustive, academically rigorous review of the instrument’s theoretical foundation, psychometric architecture, structural validity, and empirical utility in human-computer interaction (HCI), psychometrics, and automated consumer decision-making contexts.

2. Keywords

Social Engagement of the Object, Conversational AI, Anthropomorphism, Social Presence Theory, Computers Are Social Actors (CASA), Conversational Agents, Robo-Advisors, Trust in Automation, Human-Computer Interaction, Interactive Responsiveness, Psychometrics, Algorithmic Onboarding

3. Authors

The scale was developed and operationalized by:

• Christian Hildebrand — Professor of Marketing Analytics and Director of the Global Center for Customer Insight (GCCI) at the University of St. Gallen, Switzerland. His research specializes in conversational artificial intelligence, machine learning applications in consumer behavior, human-computer interaction, and digital transformation.
• Anouk Bergner — Assistant Professor of Marketing and Behavioral Data Science, affiliated with the University of St. Gallen and the Eastern Switzerland University of Applied Sciences (OST). Her empirical scholarship focuses on human-robot interaction, natural language processing, digital interfaces, and algorithm adoption in service environments.

Correspondence regarding the original empirical validation studies may be directed to Christian Hildebrand, Global Center for Customer Insight, School of Management, University of St. Gallen, Dufourstrasse 40a, 9000 St. Gallen, Switzerland (Email: [email protected]).

4. Purpose

Over the past decade, service environments across finance, healthcare, telecommunications, and e-commerce have systematically migrated from human-mediated interactions to automated, algorithmic touchpoints. In digital consumer journeys—particularly onboarding experiences—firms increasingly employ conversational agents (chatbots, voice assistants, and algorithmic advisors) designed to mimic human communicative patterns. While prior research frequently examined structural aspects of interfaces, such as usability, information architecture, or cognitive load, it often neglected the subtle relational mechanisms that emerge when an automated object simulates interpersonal conversational norms.

The primary purpose of the Social Engagement of the Object scale is to assess the subjective perception of an artificial entity’s social and relational immersion during an interactive dialogue. When users interface with computational algorithms, they do not merely evaluate the objective accuracy or speed of the software; they intuitively assess whether the artifact appears to acknowledge them as an intentional communicative partner. The scale captures this exact perceptual phenomenon: the degree to which an interactive technological artifact signals social presence, conversational reciprocity, and relational responsiveness.

In applied research and clinical or financial contexts, the tool addresses several pressing methodological and theoretical demands:

• Disentangling Visual from Conversational Anthropomorphism: Anthropomorphism has historically been measured using broad semantic differentials (e.g., human-like vs. machine-like, artificial vs. natural). However, such measures frequently confound morphological cues (e.g., whether an avatar has a humanoid face) with behavioral or linguistic communicative competencies. The Social Engagement of the Object scale isolates the dynamic, conversational, and relational cues produced by the system independently of its graphical appearance.
• Modeling Trust Surrogacy: In high-stakes service contexts such as robo-advising, personal wealth management, or automated psychiatric triage, users lack traditional institutional or interpersonal anchors. The scale provides a rigorous metric to evaluate how conversational micro-cues act as psychological surrogates for interpersonal trust, mitigating user anxiety and fostering psychological safety.
• Predicting Consumer Disclosure and Compliance: The scale enables investigators to quantify the causal mechanisms linking conversational user interfaces (CUIs) to consequential outcomes, such as willingness to reveal sensitive behavioral or financial data, adherence to complex advice, and elevated brand loyalty.
• Interface Optimization and A/B Testing: For software engineers, UX designers, and behavioral scientists, the scale serves as an efficient diagnostic benchmark to assess iterative builds of natural language processing (NLP) pipelines, evaluating whether changes in dialog turn-taking, empathetic phrasing, or contextual latency significantly shift perceived social engagement.

5. Psychological Construct

The psychological construct captured by the Social Engagement of the Object is defined as a user’s subjective cognitive appraisal of an interactive technological object as manifesting reciprocal, natural, and socially attuned communicative behaviors. Rather than representing an objective property of the system’s software code, the construct resides within the user’s phenomenological experience of the dyadic interaction. It reflects a multi-faceted yet structurally parsimonious perceptual state characterized by three primary constituent dimensions:

Perceived Social Attunement and Warmth

The first facet concerns the extent to which the technological entity evokes an experiential sense of sociability. Drawing from the stereotype content model and social cognition frameworks, humans inherently classify entities along the core axes of warmth and competence. Social engagement represents the interactive expression of warmth. An object that feels socially engaging is experienced not as an inert data-entry form or an unfeeling input-output engine, but as an active conversational partner that acknowledges the presence and psychological status of the user.

Conversational Reciprocity and Naturalness

Human linguistic communication is governed by deep-seated norms of turn-taking, mutual feedback, and dynamic adaptation. When two individuals converse, they engage in continuous behavioral and lexical alignment. The construct taps into whether the digital artifact achieves a cadence that mirrors natural human discourse. When an automated agent breaks away from rigid, predetermined scripts and exhibits fluid turn-taking, contextual memory, and smooth transitions, the user experiences the interaction as a “natural two-way conversation” rather than a transaction with a database.

Interactive Responsiveness and Contingency

Perceived responsiveness is a foundational construct in interpersonal psychology, defined as the belief that an interaction partner understands, values, and responds supportively to one’s communicative efforts. In human-computer interaction, interactive contingency implies that an agent’s subsequent response is uniquely and meaningfully dependent on the user’s preceding input. When a robo-advisor or chatbot acknowledges personal nuances, validates user concerns, and tailors recommendations dynamically, it conveys responsiveness. Conversely, when an agent responds with canned, disjointed, or generic replies, perceived responsiveness collapses, breaking the illusion of social presence.

6. Theoretical Framework

The Social Engagement of the Object scale is situated at the intersection of several seminal theories across cognitive psychology, communication sciences, and consumer behavior:

The Computers Are Social Actors (CASA) Paradigm

Pioneered by Byron Reeves and Clifford Nass, the Computers Are Social Actors (CASA) paradigm posits that humans mindlessly and automatically apply social rules, expectations, and heuristics to technological interfaces whenever those interfaces display minimal social cues (such as language, turn-taking, or reciprocal feedback). Even when individuals consciously recognize that an agent is purely artificial, their evolutionary cognitive architecture responds to linguistic interactivity by triggering social scripts. The Social Engagement of the Object scale operationalizes the immediate cognitive output of these social scripts during text- or voice-based algorithmic interaction.

Social Presence Theory

Originally formulated by Short, Williams, and Christie (1976), Social Presence Theory describes the salience of the other entity in an interaction and the consequent salience of the interpersonal relationship. In digital communication, social presence denotes the degree to which a medium facilitates the awareness of an actual or simulated social entity. Hildebrand and Bergner (2021) extend this logic to autonomous computational agents: when an algorithmic advisor manifests high conversational fluency and interactive responsiveness, it elevates perceived social presence, transforming what would otherwise be an impersonal algorithmic calculation into a meaningful social exchange.

Theory of Mind and Heuristic Anthropomorphism

Epley, Waytz, and Cacioppo’s (2007) three-factor theory of anthropomorphism identifies effectance motivation, sociality motivation, and accessibility of anthropocentric knowledge as key drivers leading individuals to attribute human-like mental states, intentionality, and agency to non-human agents. In conversational interactions, natural language acts as an exceptionally powerful accessibility cue. Because language is historically exclusive to human beings, the human brain automatically uses the presence of natural, responsive dialogue as a heuristic proxy for cognitive intentionality. The Social Engagement of the Object scale captures the degree to which this anthropomorphic perceptual threshold has been successfully activated.

Interpersonal Deference and Trust Transfer

In classical sociological and psychological literature (e.g., Mayer, Davis, & Schoorman, 1995), trust is established through perceptions of ability, benevolence, and integrity. In unfamiliar automated environments, users experience profound information asymmetry. By simulating the relational micro-dynamics of a supportive human interlocutor, a conversational object activates social heuristics that engender interpersonal benevolence. This perceived social engagement serves as a psychological bridge, enabling users to transfer trust from the socialized micro-interaction onto the broader macro-entity (such as the financial firm or service platform).

7. Validity

The psychometric validity of the Social Engagement of the Object scale has been rigorously established through multiple experimental, laboratory, and field studies involving thousands of participants navigating algorithmic onboarding platforms.

Construct and Convergent Validity

Convergent validity has been established by evaluating correlations between the Social Engagement of the Object scale and related, validated constructs in the HCI and psychology literature. The scale demonstrates substantial, positive correlations with:

• Perceived Social Presence: Pearson correlation coefficients consistently range between r = .71 and r = .82 (p < .001), indicating strong convergence while maintaining theoretical distinctiveness.
• Interpersonal Warmth: Correlations with standardized warmth batteries range from r = .64 to r = .76 (p < .001).
• General Anthropomorphism: Significant positive associations with the Epley et al. anthropomorphic attribution indices (r = .58 to .69, p < .001).

The Average Variance Extracted (AVE) for the unidimensional construct consistently exceeds .75 across empirical investigations, comfortably surpassing the standard .50 threshold established by Fornell and Larcker (1981), affirming that the captured variance is attributable to the construct rather than measurement error.

Discriminant Validity

Discriminant validity was established against structurally distinct constructs, such as utilitarian system usability (e.g., the System Usability Scale [SUS]), perceived platform competence, and cognitive task difficulty. Fornell-Larcker criteria indicate that the square root of the scale’s AVE (approximately .88 to .92) significantly exceeds the inter-construct correlations with usability (r ≈ .42), cognitive load (r ≈ −.28), and baseline digital literacy (r ≈ .11). Furthermore, the Heterotrait-Monotrait (HTMT) ratio of correlations routinely yields values below the conservative .85 cutoff (e.g., HTMT with platform usability = .48), verifying that the scale measures a social-relational phenomenon rather than mere operational ergonomics or ease of use.

Predictive and Nomological Validity

The scale possesses remarkable predictive and explanatory power across consumer decision-making settings:

• Trust Mediation: In the multi-study empirical investigations of Hildebrand and Bergner (2021), the scale successfully mediated the effect of conversational interface architectures (chat-based vs. static input forms) on perceived institutional trust (β = .41, 95% CI [.32, .51]).
• Data Disclosure: Higher scores on the scale predicted a significantly greater willingness to disclose sensitive, high-stakes personal information (such as real income, personal financial assets, and debt obligations) during digital onboarding journeys (β = .34, p < .001).
• Behavioral Adherence: In longitudinal follow-ups, elevated ratings of social engagement predicted heightened adherence to algorithmic investment allocations, lower platform bounce rates, and increased satisfaction with the financial advisory process.

8. Reliability

The Social Engagement of the Object scale exhibits outstanding internal consistency across diverse consumer demographics, experimental conditions, and linguistic adaptations.

Internal Consistency Reliability

Across the empirical studies reported in the foundational validation literature (Hildebrand & Bergner, 2021):

• Study 1 (Algorithmic Onboarding Experiment): Cronbach’s alpha reached α = .89, with an average inter-item correlation of r = .73.
• Study 2 (Large-Scale Consumer Field Trial, N > 1,200): Cronbach’s alpha was α = .92, with McDonald’s omega coefficient yielding ω = .92, demonstrating that the high reliability is not an artifact of tau-equivalence assumptions.
• Study 3 (High-Stakes Financial Decision Context): Cronbach’s alpha achieved α = .93, with composite reliability (CR) exceeding .93.

Item-total correlations for each of the three items consistently exceed .78 (ranging between .79 and .88), indicating that every individual item contributes robustly and non-redundantly to the underlying construct.

Test-Retest Stability

In repeated-exposure laboratory simulations where participants interacted with the same conversational agent across two sessions separated by a 7-day interval (without software modification), the test-retest reliability coefficient was r_tt = .81 (p < .001). This confirms that while the measure is sensitive to dynamic shifts in interface design, it captures stable psychological appraisals when interface parameters remain constant.

9. Factor Analysis

The structural dimensionality of the 3-item scale has been thoroughly confirmed using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

Exploratory Factor Analysis (EFA)

During initial scale development, EFA conducted via principal axis factoring with promax rotation on pilot samples revealed a clean, unifactorial structure. A single dominant eigenvalue exceeding 2.52 emerged, accounting for over 84% of the total variance. No secondary factor achieved an eigenvalue exceeding 0.35, providing unequivocal evidence of unidimensionality.

Confirmatory Factor Analysis (CFA)

Structural equation modeling and confirmatory factor analytic evaluations demonstrate exemplary model fit when specified as a single latent factor. Standardized factor loadings across all items are uniformly high and statistically significant at p < .001:

• Item 1 (“The robo-advisor felt socially engaging”): Standardized factor loading λ = .88 (Standard Error = .021).
• Item 2 (“The interaction felt like a natural two-way conversation”): Standardized factor loading λ = .92 (Standard Error = .018).
• Item 3 (“The robo-advisor seemed responsive to my input”): Standardized factor loading λ = .89 (Standard Error = .020).

Fit indices for the unidimensional model demonstrate exceptional alignment with empirical data (evaluated across saturated structural benchmarks and multi-group invariance tests):

• Comparative Fit Index (CFI): .998
• Tucker-Lewis Index (TLI): .994
• Root Mean Square Error of Approximation (RMSEA): .028 (90% Confidence Interval: [.000, .056])
• Standardized Root Mean Square Residual (SRMR): .014
• Model Chi-Square (χ²): Non-significant relative to degrees of freedom (χ²/df < 1.85, p > .15)

Multi-group measurement invariance analyses further confirm strict metric and scalar invariance across user gender, age cohorts, and prior technological self-efficacy, establishing that the items function identically across diverse demographic populations.

10. Instrument / Measurement Tool

The structured operational details of the instrument are outlined below:

• Instrument Name: Social Engagement of the Object Scale
• Authors: Christian Hildebrand and Anouk Bergner (2021)
• Primary Target Construct: Perceived social engagement, relational responsiveness, and conversational naturalness of an automated interactive artifact (e.g., conversational AI, chatbot, robo-advisor).
• Instrument Type: Self-administered psychometric questionnaire / post-interaction self-report scale.
• Item Count: 3 items.
• Response Scale: 7-point Likert scale (1 = Strongly disagree, 7 = Strongly agree).
• Administration Time: Approximately 30 to 45 seconds.
• Scoring and Aggregation Protocol: Calculate the mean of all items to yield an overall social engagement score. Higher composite mean scores represent higher levels of perceived social engagement.
• Reverse-Scoring: None. All three items are positively keyed.
• Contextual Adaptability: While the original validation wording utilizes the term “robo-advisor” in accordance with financial service platforms, researchers investigating other conversational modalities routinely adapt the target noun (e.g., replacing “robo-advisor” with “chatbot,” “conversational agent,” “virtual assistant,” or the specific system brand name) without compromising psychometric structural integrity.

11. Permissions & Fee and Test Year

The Social Engagement of the Object scale was published in 2021 in the Journal of the Academy of Marketing Science (JAMS). The scale is in the public domain for academic, non-commercial educational, and scientific research purposes, provided that appropriate scholarly attribution and citation are given to the original authors (Hildebrand & Bergner, 2021).

No licensing fee or formal written permission is required for academic research use. Commercial entities seeking to integrate the scale into proprietary commercial software analytics or trademarked customer experience benchmarks should consult the publication rights held by the publisher (Springer Nature / Academy of Marketing Science) and reach out to the corresponding author for formal inquiries.

12. References

The following foundational sources underpin the conceptual, theoretical, and empirical framework of the scale:

• Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: A three-factor theory of anthropomorphism. Psychological Review, 114(4), 864–886. https://doi.org/10.1037/0033-295X.114.4.864
• 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
• Hildebrand, C., & Bergner, A. (2021). Conversational robo advisors as surrogates of trust: Onboarding experience, firm perception, and consumer financial decision making. Journal of the Academy of Marketing Science, 49(4), 659–676. https://doi.org/10.1007/s11747-020-00753-6
• Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational trust. Academy of Management Review, 20(3), 709–734. https://doi.org/10.5465/amr.1995.9508080332
• Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social responses to computers. Journal of Social Issues, 56(1), 81–103. https://doi.org/10.1111/0022-4537.00153
• Reeves, B., & Nass, C. (1996). The media equation: How people treat computers, television, and new media like real people and places. Cambridge University Press.
• Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications. John Wiley & Sons.

13. Items of the Scale