Affective Response (Negative) (ARN)

1. Abstract

The Affective Response (Negative) (ARN) scale is an eight-item psychometric state instrument originally operationalized by Jiangang Du, Xiucheng Fan, and Tianjun Feng (2011) to capture the intensity, fluctuation, and valence of acute negative emotional experiences during dynamic interpersonal interactions and service encounters. Grounded in the structural traditions of the Positive and Negative Affect Schedule (PANAS) and circumplex models of human emotion, the ARN was engineered specifically to evaluate transient, real-time emotional episodes rather than stable affective dispositions or enduring traits. The instrument incorporates eight affective markers that span both high-arousal negative states (such as distress, anger, and anxiety) and moderate-to-low arousal aversive states (such as disgust and generic displeasure), providing comprehensive coverage of the unpleasant valence spectrum. Typically administered using a multi-point intensity format (ranging from 1 = “Not at all” to 7 = “Extremely”), the scale demonstrates exceptional psychometric properties, consistently exhibiting high internal consistency reliability (Cronbach’s α typically ranging between .88 and .94 across experimental measurement waves), robust composite reliability (> .90), and distinct convergent, discriminant, and criterion validity. Confirmatory factor analyses corroborate an overarching unidimensional state factor representing acute negative emotional activation, while sensitivity analyses confirm its capacity to capture temporal fluctuations across sequential interaction phases without succumbing to excessive measurement reactivity. This article provides a comprehensive academic review of the ARN scale, explicating its theoretical heritage, construct composition, psychometric evidence, structural dimensionality, administrative protocols, and research applications across consumer psychology, organizational behavior, and clinical affective science.

2. Keywords

Affective Response Negative, ARN, state affect, emotional contagion, PANAS, service encounters, negative valence, emotional arousal, psychometrics, consumer emotion, affective dynamics, scale validation.

3. Authors

The Affective Response (Negative) scale was adapted and validated in its localized eight-item temporal form by a research team investigating multi-phase emotional contagion mechanisms in interpersonal and commercial contexts:

• Jiangang Du — Professor of Marketing, Business School, Nankai University, Tianjin, China. Specialized in service marketing, consumer psychology, customer experience management, and interactive emotional dynamics.
• Xiucheng Fan — Professor of Marketing, School of Management, Fudan University, Shanghai, China. Specialized in relationship marketing, service quality evaluation, brand equity, and customer value co-creation.
• Tianjun Feng — Professor of Operations Management and Information Systems, School of Management, Fudan University, Shanghai, China. Specialized in behavioral operations management, service system design, and quantitative decision modeling.

The landmark empirical validation of the scale appeared in their foundational paper, “Multiple emotional contagions in service encounters,” published in the Journal of the Academy of Marketing Science (2011).

4. Purpose

The primary purpose of the Affective Response (Negative) (ARN) scale is to deliver a psychometrically rigorous, brief, and temporally agile measure of state negative affect that can be administered repeatedly within condensed experimental or longitudinal sessions without inducing significant participant fatigue, cognitive depletion, or methodological artifact. In human interaction studies, emotional reactions are fundamentally non-static; they emerge, intensify, attenuate, and transform across discrete transactional touchpoints. Traditional trait scales (such as the trait form of the PANAS or the Neuroticism subscales of personality inventories) assess cross-situational affective tendencies and are fundamentally insensitive to acute, event-triggered affective shocks. Conversely, long-form affective inventories (such as the 60-item PANAS-X or the Profile of Mood States) are excessively burdensome for protocols requiring repeated administrations across multiple phases of a single laboratory experiment or field trial.

The ARN bridges this methodological divide. It was developed to address critical research questions regarding how situational disruptions, interpersonal tensions, and adverse nonverbal or verbal stimuli trigger acute negative feeling states. Specifically, in service encounter research, customers frequently enter an interaction with a baseline emotional state, encounter an employee displaying genuine or fabricated emotional cues, experience service failures or recoveries, and exit with an altered affective profile. The ARN enables researchers to map these granular shifts, isolating the degree to which customer distress is attributable to direct cognitive evaluations of service performance versus involuntary emotional contagion stemming from the service provider’s expressed emotion.

Beyond commercial and consumer research, the ARN serves vital purposes in organizational psychology and clinical research. In workplace settings, the scale is employed to evaluate the affective toll of abusive supervision, customer incivility, and emotional labor on front-line employees. In clinical and experimental psychopathology, the scale functions as an effective manipulation check and momentary outcome index following laboratory stress inductions, social exclusion paradigms (e.g., Cyberball), threat exposure protocols, or cognitive bias modification tasks. By measuring the acute intensity of negative emotional responses with eight balanced descriptors, investigators can precisely quantify changes in emotional equilibrium provoked by target interventions.

5. Psychological Construct

The psychological construct evaluated by the ARN is state negative affect (NA), defined as a momentary subjective experience characterized by unpleasant emotional engagement, mental distress, and varying degrees of physiological and psychological arousal. Unlike enduring emotional traits or general neurotic tendencies, state negative affect reflects transient phenomenological responses to immediate environmental, social, or cognitive events.

Historically, the construct of negative affect has been conceptualized either as an undifferentiated general dimension of subjective distress or as a collection of discrete basic emotional states (e.g., anger, fear, sadness, disgust). The ARN operationalizes negative affect as a coherent, higher-order state construct while intentionally sampling indicators that reflect distinct emotional manifestations. This balance prevents the scale from becoming an overly narrow measure of a single discrete emotion (e.g., pure state anger) while ensuring broad coverage across the negative valence domain.

The core dimensions and emotional manifestations captured by the eight items of the ARN include:

• High-Arousal Inward-Directed Tension (Anxiety and Distress): Marked by subjective feelings of unease, apprehension, and distress. In high-arousal negative states, individuals experience rapid autonomic nervous system activation, cognitive narrowing, and hyper-vigilance toward perceived threats or service breakdown points.
• High-Arousal Outward-Directed Agitation (Hostility and Anger): Characterized by irritation, annoyance, and aggressive action tendencies. This component is particularly critical in interpersonal conflict and customer service failures, where thwarted goals or perceptions of unfairness prompt acute, externally focused negative valence.
• Aversive Satiation and Rejection (Disgust and Contempt): Marked by revulsion, deep distaste, and psychological distancing. These items capture adverse reactions provoked by unhygienic environments, offensive interpersonal conduct, or profound violations of social norms.
• Dysphoric Agitation and Jitteriness: Characterized by a state of being unsettled, perturbed, or emotionally shaken. This reflects the acute physiological restlessness that often accompanies sudden, unexpected adverse information or interpersonal awkwardness.

Importantly, the ARN emphasizes that while these items tap into phenomenologically discernible emotional nuances, they load strongly onto a shared underlying negative affective engine during high-stress interactions. When an individual encounters an aversive interpersonal experience, negative emotions rarely emerge in total isolation; rather, anger, distress, and irritation typically co-occur in an integrated, aversive affective reaction. The ARN captures the overall magnitude of this composite psychological construct with high fidelity.

6. Theoretical Framework

The structural and conceptual scaffolding of the ARN scale rests upon three primary theoretical paradigms within affective science and social psychology: the Two-Factor Model of Structure of Affect, the Circumplex Model of Emotion, and the Theory of Primitive Emotional Contagion.

The Two-Factor Model of Affect

Formulated extensively by David Watson, Lee Anna Clark, and Auke Tellegen (1988), the two-factor model posits that human affective experience is broadly organized around two orthogonal, overarching dimensions: Positive Affect (PA) and Negative Affect (NA). Within this paradigm, Positive Affect represents the extent to which an individual feels enthusiastic, active, and alert, whereas Negative Affect reflects a general dimension of subjective distress and unpleasurable engagement that subsumes a variety of aversive mood states, including anger, contempt, disgust, guilt, fear, and nervousness. Watson and Tellegen conceptualized these dimensions as biological and psychological systems that function relatively independently. The ARN draws its core semantic anchors directly from the negative pole of this two-factor structure, treating negative emotionality as a unified, broad dimension that can be acutely activated by situational stressors.

The Circumplex Model of Emotion

Complementing the two-factor perspective, James Russell’s Circumplex Model of Affect (1980) posits that affective states are distributed along a continuous two-dimensional space defined by valence (pleasure versus displeasure) and arousal (activation versus deactivation). Under this framework, affective states cannot be understood strictly by valence alone; the accompanying physiological and psychological activation level is equally determinative. The ARN deliberately selects items situated within the high-activation/displeasure quadrant (e.g., distressed, hostile, nervous) while incorporating items extending toward intermediate-activation displeasure. By spanning this conceptual continuum, the ARN ensures that respondents who express their negative affect via high-energy agitation as well as those experiencing intense subjective displeasure without hyper-arousal are accurately indexed.

Emotional Contagion Theory

The foundational application of the ARN by Du, Fan, and Feng (2011) was contextualized within Elaine Hatfield, John Cacioppo, and Richard Rapson’s (1994) Theory of Primitive Emotional Contagion. This theory posits that human beings possess an automatic, non-conscious tendency to mimic and synchronize their facial expressions, vocalizations, postures, and movements with those of another person, converging emotionally as a result. Du and colleagues advanced this framework into service settings, demonstrating that emotional contagion does not occur merely as a single static transfer; rather, multiple emotional contagions unfold across an interaction. As employees display positive or negative emotional cues across successive interaction phases, customers’ internal affective balance fluctuates. The ARN provides the quantitative mechanism required to track these psychological dynamics, capturing the rapid ignition of negative affect when customers are exposed to an employee’s negative display or interpersonal disinterest.

7. Validity

The validity of the ARN has been substantiated through extensive psychometric testing across multiple empirical studies, involving experimental laboratory designs, field studies, and structural equation modeling (SEM).

Construct and Convergent Validity

Convergent validity is established when indicators of a theoretical construct share a high proportion of common variance. In structural equation modeling conducted by Du et al. (2011), all eight items of the ARN demonstrated high, statistically significant factor loadings on the latent negative affect construct, with completely standardized loadings ranging from .72 to .91 (all p < .001). Furthermore, the Average Variance Extracted (AVE) consistently surpassed the established psychometric benchmark of .50 (Fornell & Larcker, 1981), indicating that the latent construct explains greater variance in the observed items than is attributable to measurement error. When compared simultaneously with standardized measures of emotional valence using semantic differential scales (e.g., pleasant-unpleasant, pleased-annoyed), the ARN correlates strongly in the predicted positive direction with unpleasantness markers (r > .75, p < .001).

Discriminant Validity

Discriminant validity demonstrates that the ARN is empirically distinct from other theoretically related but conceptually separate constructs. In rigorous testing:

• Distinction from Positive Affect (ARN vs. ARP): Latent modeling shows that the ARN correlates negatively or orthogonally with the Affective Response (Positive) scale (typically r = -.32 to -.55, depending on the severity of the service failure). The square root of the AVE for the ARN is significantly greater than its inter-construct correlations with positive affect, satisfying the Fornell-Larcker criterion and the Heterotrait-Monotrait ratio of correlations (HTMT < .85).
• Distinction from Cognitive Service Evaluations: Research confirms that state negative affect measured via the ARN is psychometrically distinct from cognitive evaluations of service quality, fair treatment, and core performance expectations. While negative affect is intensified by perceived service failure, confirmatory factor analyses comparing a one-factor model (combining cognitive dissatisfaction and ARN) against a two-factor model demonstrate overwhelmingly superior fit for the two-factor specification (Δχ²(1) > 120.0, p < .0001).
• Distinction from Trait Neuroticism: When administered in baseline versus post-stressor phases, the ARN demonstrates high situational variance and responsiveness, whereas trait measures remain flat, proving that the ARN measures a state reaction rather than an enduring personality disposition.

Criterion and Predictive Validity

The predictive utility of the ARN has been repeatedly confirmed through its downstream effects on consumer and organizational behavior. Empirical findings demonstrate that elevated ARN scores significantly predict:

• Decreased overall customer satisfaction and diminished repurchase intentions (β = -.38 to -.52, p < .001).
• Increased intention to engage in negative word-of-mouth (NWOM) communication (β = .44, p < .001).
• Heightened third-party complaint behavior and customer retaliatory intentions.
• Compromised relational trust and perceived corporate competence in service recovery contexts.

8. Reliability

The reliability of the ARN has been thoroughly established through metrics of internal consistency, composite reliability, and measurement stability across repeated waves.

Internal Consistency

Across empirical studies employing the ARN, the instrument demonstrates exemplary internal consistency. In the foundational work by Du, Fan, and Feng (2011), Cronbach’s alpha (α) was computed across discrete temporal measurement points within the same experimental encounter (e.g., Time 1: baseline post-arrival; Time 2: post-core service interaction; Time 3: post-encounter resolution):

• Measurement Wave 1 (T1): α = .89
• Measurement Wave 2 (T2): α = .92
• Measurement Wave 3 (T3): α = .94

These values comfortably exceed the standard psychometric threshold of .70 recommended for exploratory research and .80 for established instruments. Deletion of any individual item does not lead to an elevation in the overall scale alpha, confirming that all eight indicators contribute meaningfully and harmoniously to the composite score.

Composite Reliability and Variance Extraction

Because Cronbach’s alpha assumes tau-equivalence (equal factor loadings across items) and may consequently underestimate or overestimate true reliability, structural equation modeling researchers regularly compute Composite Reliability (CR). Across diverse experimental samples, the CR of the ARN consistently ranges between .91 and .95, while the Average Variance Extracted (AVE) typically ranges between .58 and .68. These indices confirm that measurement error accounts for less than one-third of the total variance observed in the scale items.

Test-Retest Reliability and State Sensitivity

In psychometrics, the evaluation of test-retest reliability for a state measure differs fundamentally from that of a trait measure. A state measure must exhibit stability in the absence of an intervention, yet demonstrate pronounced sensitivity and responsiveness when an emotional shock or mood induction occurs. In control conditions characterized by neutral, uninterrupted stimuli, the ARN demonstrates strong short-term stability over a 15-to-30 minute interval (test-retest r = .81 to .86). However, when an aversive stimulus (e.g., a deliberate service error or an encounter with a hostile confederate) is introduced, the ARN scores register immediate, statistically significant increases (Cohen’s d > 1.10), proving that the scale captures dynamic shifts in affective states rather than invariant response habits.

9. Factor Analysis

Extensive exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) have confirmed the unidimensional, single-factor latent structure of the eight-item ARN when measuring the state negative affect construct.

Exploratory Factor Analysis (EFA)

During initial scale development and psychometric refinement, principal component and maximum likelihood factor analyses with oblique rotations (e.g., Promax) consistently extract a single dominant eigenvalue exceeding 1.0 (Kaiser criterion). The primary factor typically accounts for 58% to 68% of the total variance across items. The scree plot clearly demonstrates a severe drop-off after the first extracted factor, with the second factor yielding an eigenvalue well below 0.8. All eight items demonstrate primary factor loadings exceeding .65, with zero significant cross-loadings, demonstrating clean unidimensional convergence.

Confirmatory Factor Analysis (CFA)

To rigorously test the structural validity of the single-factor specification, confirmatory factor analyses are evaluated using robust maximum likelihood estimation. Standard structural equation models specify the eight observed items loading directly onto a single first-order latent variable (Negative Affect), with measurement error terms constrained to be uncorrelated.

The resulting goodness-of-fit statistics across representative empirical data sets consistently indicate excellent model fit:

• Chi-Square to Degrees of Freedom Ratio (χ²/df): Typically between 1.45 and 2.30, well within the conventional standard of ≤ 3.0.
• Comparative Fit Index (CFI): Consistently ≥ .96 (often > .98).
• Tucker-Lewis Index (TLI): Consistently ≥ .95.
• Root Mean Square Error of Approximation (RMSEA): Ranges between .038 and .059, with the 90% confidence interval remaining strictly below the .08 threshold for acceptable fit.
• Standardized Root Mean Square Residual (SRMR): Consistently ≤ .035.

Item Parameter Estimates and Factor Loadings

Standardized factor loadings (λ) for the eight items across confirmed models are presented below, demonstrating high individual item reliability:

Multi-group CFA testing confirms measurement invariance across gender, participant age cohorts, and experimental conditions (configural, metric, and scalar invariance established with ΔCFI < .01 and ΔRMSEA < .015 across nested comparisons).

10. Instrument / Measurement Tool

The Affective Response (Negative) scale is a structured, self-report psychological measurement instrument designed for rapid pencil-and-paper or computer-assisted administration. Below is the technical specification of the instrument:

• Construct Assessed: Acute state negative affect (subjective negative emotional activation).
• Instrument Type: Self-administered psychometric rating scale / affective inventory.
• Number of Items: 8 distinct affective descriptors.
• Administration Modality: Online survey platforms (Qualtrics, REDCap), mobile momentary assessments (EMA), or in-person laboratory testing.
• Target Completion Time: Approximately 45 to 90 seconds, making it suitable for multi-wave experimental designs.
• Prompt Instructions: Participants are instructed: “Indicate to what extent you feel this way right now, that is, at the present moment.”
• Response Scale: Typically formatted as a 7-point Likert-type intensity continuum:
  • 1 = Not at all / Very slightly
  • 2 = A little
  • 3 = Moderately
  • 4 = Quite a bit
  • 5 = Considerably
  • 6 = Very strongly
  • 7 = Extremely

  (Note: Studies requiring synchronization with 5-point PANAS protocols may use a 1-to-5 scale anchored from 1 = “Very slightly or not at all” to 5 = “Extremely”).

• Scoring Protocol:
  • All 8 items are directly worded and positively keyed toward negative affect; therefore, no reverse-coded items exist within the instrument.
  • The overall composite ARN score is computed either as the arithmetic mean or the sum of all eight items.
  • Arithmetic Mean Calculation: ARN Score = (∑ Item Scores) / 8, yielding an interpretive scale ranging from 1.00 to 7.00.
  • Higher numerical scores correspond to greater levels of acute negative emotional distress and physiological/affective agitation.

11. Permissions & Fee and Test Year

The localized eight-item operationalization of the Affective Response (Negative) scale was published by Jiangang Du, Xiucheng Fan, and Tianjun Feng in 2011 in the Journal of the Academy of Marketing Science (Springer). Because the scale adapts public-domain and widely published lexical markers from the classical PANAS tradition (Watson, Clark, & Tellegen, 1988), it is generally classified as an open-access psychometric instrument for academic, non-commercial, educational, and scientific research purposes.

Key administrative and copyright guidelines include:

• Commercial Applications: Commercial organizations, market research firms, or corporate entities incorporating the scale into proprietary employee monitoring software, commercial customer satisfaction trackers, or fee-generating diagnostic platforms should review relevant copyright policies and consult the authors or the publisher (Springer / Academy of Marketing Science) regarding institutional permissions.
• Academic Research: Independent scholars, university researchers, and graduate students may utilize the scale in scientific inquiries free of royalty fees, provided appropriate scholarly attribution is accorded via standard academic citation of the foundational 2011 paper.
• Adaptation and Translation: Scholars translating the scale into languages other than English or Chinese are required to conduct formal forward-and-back translation procedures and verify psychometric invariance and structural fidelity.

12. References

The following foundational sources document the theoretical roots, psychometric validation, and direct applications of the Affective Response (Negative) scale:

• Du, J., Fan, X., & Feng, T. (2011). Multiple emotional contagions in service encounters. Journal of the Academy of Marketing Science, 39(3), 449–466. https://doi.org/10.1007/s11747-010-0211-1
• 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
• Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emotional contagion. Cambridge University Press. https://doi.org/10.1017/CBO9781139174138
• Lazarus, R. S. (1991). Emotion and adaptation. Oxford University Press.
• Pugh, S. D. (2001). Service with a smile: Emotional contagion in the service encounter. Academy of Management Journal, 44(5), 1018–1027. https://doi.org/10.5465/3069445
• Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161–1178. https://doi.org/10.1037/h0077714
• Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. https://doi.org/10.1037/0022-3514.54.6.1063

13. Items of the Scale