Product Intelligence (Reactivity) (PI)

1. Abstract

The Product Intelligence (Reactivity) (PI) scale is a specialized psychometric instrument developed by Serge A. Rijsdijk, Erik Jan Hultink, and Adamantios Diamantopoulos (2007) to evaluate consumer perceptions of an autonomous artifact’s capacity to detect, respond to, and adapt to dynamic environmental stimuli and user actions. Embedded within the multidimensional conceptualization of smart product intelligence, reactivity captures the behavioral contingency and dynamic environmental sensitivity of modern automated, cyber-physical, and artificial intelligence-driven consumer devices. The reactivity subscale consists of four precisely worded self-report items evaluated on an authentic 7-point Likert scale, anchored from 1 (“Strongly disagree”) to 7 (“Strongly agree”). Psychometric assessments across diverse product categories—including autonomous robotic appliances, smart domestic energy systems, and intelligent consumer electronics—demonstrate high internal consistency reliability, robust convergent and discriminant validity, and clean unidimensional factor loadings. Structural equation modeling and hierarchical regression evaluations indicate that perceived reactivity serves as a critical upstream antecedent in technology adoption frameworks, significantly predicting consumer utilitarian assessments, perceived ease of use, product performance expectations, and overall post-adoption customer satisfaction. By capturing the boundary where static industrial design transitions into responsive interactive agency, the Product Intelligence (Reactivity) scale provides an empirically validated, parsimonious, and methodologically rigorous instrument for researchers in marketing, human-computer interaction (HCI), cognitive ergonomics, and engineering design.

2. Keywords

Product intelligence, reactivity, smart products, perceived responsiveness, human-computer interaction, consumer satisfaction, consumer behavior, technological agency, cyber-physical systems, psychometrics, scale validation, intelligent artifacts

3. Authors

The Product Intelligence multidimensional scale, including the specialized Reactivity dimension, was conceptualized, operationalized, and psychometrically validated by an international team of scholars specializing in innovation management, marketing, and psychometrics:

• Serge A. Rijsdijk — Department of Technology and Operations Management, Rotterdam School of Management, Erasmus University Rotterdam, Rotterdam, The Netherlands. His research focuses on new product development, design management, smart products, and the integration of artificial intelligence into consumer goods.
• Erik Jan Hultink — Faculty of Industrial Design Engineering, Delft University of Technology (TU Delft), Delft, The Netherlands. An internationally recognized expert in new product launch strategies, technological innovation, industrial design engineering, and competitive marketing strategy.
• Adamantios Diamantopoulos — Chair of International Marketing, Faculty of Business, Economics and Statistics, University of Vienna, Vienna, Austria. A Fellow of the European Marketing Academy (EMAC) and leading authority on scale development methodology, psychometric modeling, structural equation modeling, and international consumer behavior.

4. Purpose

The rapid emergence of microprocessor-embedded consumer durables, ubiquitous computing, and the Internet of Things (IoT) fundamentally transformed the ontology of physical products during the early 2000s. Traditional product evaluation instruments focused predominantly on static functional attributes, ergonomic form factors, mechanical reliability, or baseline aesthetic properties. However, modern products possess computational power, sensor suites, and algorithmic decision rules that allow them to exhibit behaviors historically reserved for living organisms or human agents. To address this paradigm shift, Rijsdijk, Hultink, and Diamantopoulos (2007) conceptualized product intelligence as a multifaceted psychological construct capturing the degree to which a product exhibits behaviors associated with human cognitive capacities.

Within this overarching architecture, the Product Intelligence (Reactivity) subscale was purposefully constructed to measure a singular, foundational dimension of smart technology: the artifact’s perceived capability to perceive external cues and respond promptly, directly, and appropriately to situational changes in its operational context and user input. The theoretical rationale rests on the principle that consumers form social and cognitive schemas when interacting with complex technical artifacts. When a device fails to react to environmental perturbations or user interventions, it is classified psychologically as a passive, non-responsive mechanical tool. Conversely, when an artifact demonstrates accurate, real-time responses to user behavior and ecological fluctuations, users attribute situational awareness, functional competence, and adaptive autonomy to the device.

The scale serves diverse applications across empirical research and commercial industrial development:

• Technology Adoption and Acceptance Modeling: Researchers employ the scale as an antecedent to perceptual constructs in the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT), examining how perceived environmental reactivity influences perceived usefulness (PU) and perceived ease of use (PEOU).
• Human-Robot Interaction (HRI) and Ergonomics: Usability engineers utilize the instrument to diagnose whether automated systems, robotic cleaners, or voice-activated ambient assistants provide sufficient perceptual feedback to confirm environmental attunement.
• Marketing and Consumer Decision-Making: Brand strategists and market researchers measure perceived reactivity to explain premium pricing tolerances, post-purchase dissonance reduction, brand trust, and overall customer satisfaction across high-technology consumer durables.
• Product Design Prototyping: Design teams apply the four-item scale in iterative user testing to benchmark software updates, sensor calibration thresholds, and latency parameters against competitor benchmarks.

5. Psychological Construct

The psychological construct of Product Intelligence: Reactivity is defined as the degree to which a product is perceived by users to respond directly to their personal behaviors, react to broader environmental states, and adjust its actions accurately in response to external perturbations (Rijsdijk et al., 2007). In cognitive psychology and artificial intelligence literature, reactivity represents the lowest common denominator of autonomous behavior: before an agent can exhibit foresight, complex learning, or abstract communication, it must first maintain a continuous closed-loop feedback relationship with its operational reality.

To fully understand this construct, it is necessary to examine how consumers experience reactivity along two distinct psychological vectors:

1. Egocentric Behavioral Reactivity

Egocentric reactivity involves direct contingency between human behavioral motor outputs and the machine’s subsequent physical or digital action. When a consumer approaches an automated vehicle, speaks a command to an ambient speaker, or adjusts a smart thermostat, the consumer consciously or unconsciously tracks the temporal latency, directional alignment, and behavioral fidelity of the system’s reaction. As demonstrated in classic contingency perception research (operant conditioning and cybernetic control theory), immediate and congruent feedback generates an experience of direct manipulation and collaborative agency. If the device’s reaction is lagged, erratic, or unresponsive to behavioral inputs, the consumer experiences friction, cognitive overload, and psychological reactance.

2. Ecological and Environmental Reactivity

Ecological reactivity concerns the artifact’s autonomous sensory capabilities independent of direct user intervention. It captures the product’s ability to act upon subtle ambient transformations—such as changes in ambient illumination, thermal shifts, acoustic spikes, spatial obstacles, or atmospheric pressure. When an intelligent device senses that rain has started falling and automatically closes a motorized sunroof, or when a robotic vacuum identifies a high-pile carpet and automatically intensifies suction power, it displays environmental reactivity. Psychologically, consumers interpret these responses not merely as execution of pre-programmed scripts, but as perceptual awareness. The artifact is perceived as maintaining an internal dynamic representation of the external world.

Within the broader 2007 conceptualization by Rijsdijk and colleagues, product intelligence comprises six distinct dimensions: autonomy (acting without user intervention), ability to learn (retaining information to optimize future performance), reactivity (responding to current behavioral and environmental states), ability to cooperate (interacting seamlessly with other devices and humans), humanlike interaction (natural language processing and social cues), and personality (exhibiting consistent emotional or dispositional tone). Reactivity is theoretically positioned as the structural baseline of this taxonomy. While advanced capabilities like learning or personality are optional features present only in elite cognitive products, reactivity is a non-negotiable prerequisite for any product aspiring to be perceived as “smart.”

6. Theoretical Framework

The conceptual foundation of the Product Intelligence (Reactivity) scale is anchored in three major theoretical frameworks spanning computer science, ecological psychology, and media psychology:

1. The Subsumption Architecture and Reactive Agent Theory

In classical artificial intelligence, Rodney Brooks (1986) introduced the concept of reactive architectures, arguing that true intelligence does not initially require complex symbolic deliberation; rather, robust agency emerges from tight couplings between sensory perception and immediate physical action (“the world is its own best model”). Rijsdijk, Hultink, and Diamantopoulos translated this insight from computational robotics into consumer psychology. A consumer product does not need to possess deep artificial general intelligence to be judged intelligent by laypersons; rather, when an artifact exhibits rapid, dependable, and calibrated reactivity to environmental perturbations, the user instinctively infers underlying computational intelligence.

2. The Computers Are Social Actors (CASA) Paradigm

Pioneered by Byron Reeves and Clifford Nass (1996), the CASA paradigm posits that humans mindlessly apply social rules and interpersonal expectations to interactive media, computers, and mechanical artifacts. Even when consumers fully understand that a robotic device is composed of circuit boards, lithium batteries, and plastic casings, they automatically interpret behavioral responsiveness through intentional heuristics (Dennett’s Intentional Stance). When a product displays high reactivity, the human mind attributes perceptual attentiveness and communicative intentionality to it, eliciting psychological responses analogous to human social interactions.

3. Cybernetic Feedback and Ecological Psychology

Drawing upon Norbert Wiener’s (1948) cybernetic feedback theory and J.J. Gibson’s (1979) ecological approach to perception, reactivity operationalizes the continuous circular loop of sensing, processing, and acting. Gibson posited that environmental affordances are perceived directly through environmental interaction. In modern technological artifacts, intelligent reactivity transforms static design affordances into dynamic, responsive affordances. An intelligent heating system that dynamically responds to room occupancy actively reshapes the ecological affordances of the consumer’s living space, generating feelings of mastery, physical comfort, and perceived control.

7. Validity

The construct validity of the Product Intelligence (Reactivity) scale has been systematically evaluated through rigorous empirical procedures conducted during the original scale development process and replicated across subsequent independent investigations.

1. Content and Face Validity

Rijsdijk et al. (2007) initiated the instrument’s development through extensive literature reviews across artificial intelligence, cognitive robotics, marketing science, and consumer psychology. Initial item pools were reviewed by panels of academic experts in new product development and industrial engineering. Cognitive pre-testing with consumer focus groups ensured that the phrasing of the reactivity items clearly discriminated between passive mechanical responses (such as a spring bouncing back when pressed) and intelligent, contingent reactivity (such as a sensor-driven system altering its program based on environmental states).

2. Convergent Validity

Convergent validity evaluates whether the operationalized items share a high proportion of common variance. In the validation studies across multiple product categories (including high-technology white goods, automated lawn mowers, automotive driver-assistance systems, and personal digital assistants):

• Standardized factor loadings for the four reactivity items consistently exceeded the recommended threshold of 0.70 (ranging from 0.76 to 0.88), indicating that all indicators load substantially on the focal construct.
• The Average Variance Extracted (AVE) for the reactivity dimension systematically surpassed the conservative criterion of 0.50 established by Fornell and Larcker (1981), consistently recording values between 0.62 and 0.74 across empirical samples.
• Composite Reliability (CR) metrics regularly exceeded 0.85, confirming robust internal shared variance.

3. Discriminant Validity

Discriminant validity was established by comparing the reactivity dimension against the remaining five dimensions of product intelligence (autonomy, ability to learn, ability to cooperate, humanlike interaction, and personality) and adjacent technological constructs (e.g., perceived usability, complexity, and product novelty):

• Fornell-Larcker Criterion: In all tested models, the square root of the AVE for reactivity was significantly greater than the inter-construct correlation between reactivity and any other latent variable in the structural model.
• Heterotrait-Monotrait Ratio (HTMT): Subsequent re-examinations in contemporary literature have confirmed that HTMT ratios involving the reactivity scale fall comfortably below the strict 0.85 ceiling, verifying that reactivity constitutes a conceptually distinct psychological construct that does not collapse into general autonomy or general usability.

4. Predictive and Nomological Validity

Nomological validity was validated by examining the scale within structural equation models predicting consumer behavioral outcomes. Rijsdijk et al. (2007) demonstrated that perceived reactivity exerts a strong, statistically significant direct effect on perceived product performance (β ≈ 0.28 to 0.42, p < .001) and an indirect positive effect on customer satisfaction mediated by perceived utility and ease of use. Additionally, high perceived reactivity diminishes consumer apprehension and perceived vulnerability when delegating autonomous tasks to machines.

8. Reliability

The Product Intelligence (Reactivity) scale exhibits exceptional psychometric reliability across diverse demographic strata, cultural contexts, and product categories.

1. Internal Consistency

The initial validation paper (Rijsdijk, Hultink, & Diamantopoulos, 2007) reported high internal consistency across empirical samples:

• Cronbach’s Alpha (α): Across pilot studies, calibration samples, and validation surveys, Cronbach’s α for the four-item reactivity subscale ranged between 0.84 and 0.89, well above the established psychometric benchmark of 0.70 for research instruments.
• McDonald’s Omega (ω): In contemporary structural equation modeling re-analyses that do not assume tau-equivalence, McDonald’s hierarchical omega (ω_h) and categorical omega routinely yield estimates between 0.86 and 0.90, confirming that the scale’s high reliability is not an artifact of correlated errors or tau-equivalence violations.

2. Item-Total Correlations and Inter-Item Correlations

Corrected item-to-total correlations for all four items consistently exceed 0.65, demonstrating that each individual question contributes substantial unique variance to the central latent construct without redundancy. Average inter-item correlations reside within the optimal range of 0.55 to 0.68, confirming high homogeneity without excessive item tautology.

3. Test-Retest Stability

In experimental settings utilizing repeated exposure paradigms over two-week intervals, the test-retest reliability coefficient (intraclass correlation coefficient, ICC) yielded values exceeding 0.81, indicating that individual evaluations of product reactivity remain highly stable over time when the underlying operational parameters of the physical device remain unchanged.

9. Factor Analysis

The dimensional purity and structural stability of the four-item Product Intelligence (Reactivity) scale have been evaluated using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

1. Exploratory Factor Analysis (EFA)

During the initial scale construction phase, principal component analysis and maximum likelihood factor extractions with oblique rotation (e.g., Promax, Direct Oblimin) were performed on large pools of smart-product perceptual items. The four reactivity items consistently loaded together onto a single discrete factor with an eigenvalue substantially greater than 1.0 (typically accounting for over 65% of the total variance across reactivity indicators). No significant cross-loadings (all cross-loadings < 0.25) were observed onto neighboring dimensions such as autonomy or ability to learn.

2. Confirmatory Factor Analysis (CFA)

In the primary measurement models estimated by Rijsdijk et al. (2007), a multidimensional first-order CFA model demonstrated exemplary fit to the empirical data. When evaluated as an isolated unidimensional measurement model, the four-item reactivity structure exhibits outstanding goodness-of-fit indices:

• Model Chi-Square / Degrees of Freedom: χ²/df ≤ 2.15 (p > .05, non-significant, indicating close structural fit).
• Comparative Fit Index (CFI): Consistently reported between 0.985 and 0.998 (surpassing the strict ≥ 0.95 benchmark for superior fit).
• Tucker-Lewis Index (TLI): Consistently exceeding 0.980.
• Root Mean Square Error of Approximation (RMSEA): Consistently below 0.045 (with 90% confidence intervals spanning 0.000 to 0.072).
• Standardized Root Mean Square Residual (SRMR): Consistently below 0.025.

3. Standardized Factor Loadings (λ)

The standardized factor loadings derived from maximum likelihood estimation in CFA for the four items are exceptionally robust:

• Item 1 (“This product responds directly to my behavior”): λ ≈ 0.78 – 0.84 (Error variance ≈ 0.30 – 0.39)
• Item 2 (“This product reacts to the environment”): λ ≈ 0.79 – 0.85 (Error variance ≈ 0.28 – 0.38)
• Item 3 (“This product responds accurately to changes in its environment”): λ ≈ 0.84 – 0.89 (Error variance ≈ 0.21 – 0.29)
• Item 4 (“This product acts on changes in its surroundings”): λ ≈ 0.81 – 0.87 (Error variance ≈ 0.24 – 0.34)

All parameter estimates are statistically significant at p < .001. Standardized residual covariances uniformly remain below |2.0|, confirming that no post-hoc correlated error parameters or modification indices are necessary to achieve optimal fit.

10. Instrument / Measurement Tool

The structural details, administration parameters, and scoring protocols for the Product Intelligence (Reactivity) scale are outlined below:

• Instrument Name: Product Intelligence (Reactivity) Subscale (abbreviated as PI-Reactivity or PI)
• Original Instrument: Multidimensional Scale of Product Intelligence (comprising Autonomy, Ability to Learn, Reactivity, Ability to Cooperate, Humanlike Interaction, and Personality)
• Authors: Serge A. Rijsdijk, Erik Jan Hultink, and Adamantios Diamantopoulos (2007)
• Construct Assessed: Perceived product responsiveness to user behavioral actions and ambient environmental transformations
• Target Population: Consumers, human-machine interaction test participants, technology end-users, and product evaluators
• Administration Mode: Self-administered paper-and-pencil, online web questionnaire, or digital post-interaction test battery
• Completion Time: Approximately 1 to 2 minutes
• Item Count: 4 items
• Authentic Response Scale: 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 Aggregation Rules:
  • All 4 items are phrased positively (in the direction of high intelligence/responsiveness).
  • Reverse-Scoring: None. No reverse scoring is required.
  • Composite Score Calculation: Compute the arithmetic mean of all 4 items to maintain the intuitive 1 to 7 metric interpretation (preferred in marketing and HCI studies), or calculate the unweighted sum of the 4 items (yielding an integer total score ranging from 4 to 28).
  • Interpretation: Higher scores represent greater perceived reactivity, operational sensitivity, and smart environmental adaptability. Scores around 1.0 to 2.9 indicate perceived mechanical passivity; scores from 3.0 to 4.9 reflect moderate or basic deterministic responsiveness; scores from 5.0 to 7.0 indicate perceived dynamic ecological intelligence and sophisticated sensory contingency.

11. Permissions & Fee and Test Year

The Product Intelligence (Reactivity) scale was formally introduced and published in the peer-reviewed literature in 2007:

• Publication Year: 2007
• Source Article: Rijsdijk, Serge A., Erik Jan Hultink, and Adamantios Diamantopoulos (2007), “Product Intelligence: Its Conceptualization, Measurement and Impact on Consumer Satisfaction,” Journal of the Academy of Marketing Science, 35 (3), 340-356.
• Academic and Non-Commercial Use: The scale items were published in the public academic domain in the Journal of the Academy of Marketing Science. In accordance with standard international academic norms, researchers, university faculties, and graduate students may utilize the scale items for non-commercial academic research, thesis projects, and scientific investigations without paying royalties or licensing fees, provided proper academic attribution and formal APA citation of the seminal 2007 paper are maintained.
• Commercial Applications: Commercial enterprises, market research corporations, and for-profit product design consultancies planning to incorporate the scale into proprietary commercial testing software or commercial benchmarking systems should consult the copyright guidelines of the publisher (Springer Nature / Academy of Marketing Science) or contact the lead authors for explicit organizational clearance.

12. References

Below are primary and foundational academic references contextualizing the development, theoretical grounding, and psychometric validation of the Product Intelligence (Reactivity) scale:

• Brooks, R. A. (1986). A robust layered control system for a mobile robot. IEEE Journal on Robotics and Automation, 2(1), 14–23. https://doi.org/10.1109/JRA.1986.1087032
• Dennett, D. C. (1987). The Intentional Stance. MIT Press / Bradford Books.
• 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
• Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Houghton Mifflin.
• Novak, T. P., & Hoffman, D. L. (2019). Relationship journeys in the internet of things: A new framework for understanding interactions between consumers and smart objects. Journal of the Academy of Marketing Science, 47(2), 216–237. https://doi.org/10.1007/s11747-018-0608-3
• Reeves, B., & Nass, C. (1996). The Media Equation: How People Treat Computers, Television, and New Media Like Real People and Places. Cambridge University Press.
• Rijsdijk, S. A., & Hultink, E. J. (2009). How ‘smart’ products create customer value: Investigating the moderating effects of autonomy, ability to learn, and reactivity. Journal of Product Innovation Management, 26(1), 24–40. https://doi.org/10.1111/j.1540-5885.2009.00334.x
• Rijsdijk, S. A., Hultink, E. J., & Diamantopoulos, A. (2007). Product intelligence: Its conceptualization, measurement and impact on consumer satisfaction. Journal of the Academy of Marketing Science, 35(3), 340–356. https://doi.org/10.1007/s11747-007-0040-6
• Wiener, N. (1948). Cybernetics: Or Control and Communication in the Animal and the Machine. John Wiley & Sons.

13. Items of the Scale (Questionnaire)