Post-Recommendation Product Attitude

Abstract

The Post-Recommendation Product Attitude scale is a psychometric instrument designed to measure a consumer’s evaluative judgment, cognitive appraisal, and overall valence toward a product immediately following a recommendation from an advisor, salesperson, or recommendation agent. Developed and operationalized by Jennifer K. D’Angelo and Francesca Valsesia (2023) in the Journal of Marketing Research, the scale captures how consumer perceptions are systematically shaped by the structure and delivery of advice—specifically examining the contrast between singular item recommendations and combinatory recommendations. The scale is a unidimensional, three-item self-report measure administered on a 7-point semantic differential format anchored by bipolar evaluative adjectives (Negative/Positive, Unfavorable/Favorable, Bad/Good). Psychometrically, the instrument exhibits exceptional internal consistency, with Cronbach’s alpha coefficients routinely exceeding .90 across diverse consumption domains, ranging from aesthetic goods (such as fine art) to hedonic fast-moving consumer products (such as artisanal cookies). Confirmatory factor analyses establish robust unidimensionality, with high factor loadings (λ > .85) and optimal goodness-of-fit statistics across experimental samples. The instrument displays strong convergent validity with downstream behavioral indicators, including purchase intent, willingness to pay, and downstream choice, as well as discriminant validity from perceived advisor warmth, general price sensitivity, and baseline brand familiarity. By providing a brief, psychometrically sound measure, the scale enables researchers and marketing practitioners to isolate the immediate impact of recommendation architecture, interpersonal advice, and artificial intelligence-driven recommendation systems on post-encounter consumer attitudes.

Keywords

Post-Recommendation Product Attitude, Consumer Attitudes, Semantic Differential Scale, Advice Taking, Combinatory Recommendations, Signaling Theory, Source Credibility, Evaluative Judgment, Marketing Psychometrics, Consumer Decision Making

Authors

The Post-Recommendation Product Attitude measure was established and empirically validated by:

• Jennifer K. D’Angelo — Assistant Professor of Marketing, Neeley School of Business, Texas Christian University, Fort Worth, TX, USA. Her research focuses on consumer behavior, interpersonal dynamics in retail, and the psychology of recommendation systems.
• Francesca Valsesia — Assistant Professor of Marketing, Michael G. Foster School of Business, University of Washington, Seattle, WA, USA. Her research investigates social influence, word of mouth, signaling dynamics, and the psychological mechanisms governing advisory interactions.

Purpose

The primary purpose of the Post-Recommendation Product Attitude scale is to assess the immediate, valenced evaluative response of a consumer toward a specific offering after receiving an explicit recommendation from an external source. In consumer psychology and decision sciences, the post-recommendation phase represents a critical psychological juncture wherein an individual integrates external advisory input with their own preferences, epistemic assessments of the advisor, and product attributes.

In retail, hospitality, e-commerce, and financial planning, consumers regularly encounter advisory interactions. These may originate from professional human consultants (such as art curatorial staff, sommeliers, financial planners, or retail sales associates) or algorithmic recommendation engines (such as collaborative filtering algorithms or interactive generative AI agents). While traditional attitude scales often capture generalized brand equity or enduring product loyalty over extended periods, the Post-Recommendation Product Attitude scale is designed to isolate the state-based, immediate evaluative shift elicited by the recommendation episode itself.

D’Angelo and Valsesia (2023) developed and deployed this measure to test hypotheses derived from signaling theory. Specifically, their research demonstrated that offering “combinatory recommendations” (suggesting that two complementary products be consumed or experienced together, such as pairing a specific painting with another artwork, or pairing a cookie with a specific tea) signals elevated advisor expertise. This signal leads consumers to infer deeper domain knowledge, which directly enhances the consumer’s post-recommendation attitude toward the focal product. The scale serves as the primary dependent criterion in experimental paradigms seeking to assess:

• The comparative effectiveness of various recommendation structures (e.g., single-item suggestions vs. bundled or paired suggestions).
• The downstream psychological impact of perceived advisor expertise, trustworthiness, and epistemic authority.
• The boundary conditions under which advice improves or degrades product perception, such as domains characterized by high versus low subjective taste.
• The efficacy of algorithmic versus human advice framing in digital consumption environments.

Psychological Construct

The psychological construct assessed by this instrument is evaluative product attitude formed contingent upon an advisory event. In psychometric theory, an attitude is classically defined as a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor (Eagly & Chaiken, 1993). Within the triadic model of attitudes (affect, cognition, and behavior), post-recommendation attitudes reflect a consolidated evaluative summary that synthesizes both cognitive appraisals of product utility and affective feelings of attraction or aversion generated by the social context of the advice.

The construct is unidimensional and captures the global valence dimension identified in Charles Osgood’s classic work on semantic differential structures (Osgood, Suci, & Tannenbaum, 1957). Osgood demonstrated that human evaluation of concepts across cultural and linguistic domains is dominated by an “Evaluation” factor (good–bad, positive–negative, favorable–unfavorable), which accounts for the largest proportion of variance in semantic judgments, superseding potency and activity. In the context of post-recommendation psychology, the construct represents three central facets of evaluation:

• General Valence (Negative / Positive): Reflects the directional valence of the consumer’s mental representation. A positive post-recommendation valence indicates that the advisor’s framing has activated supportive associative nodes within memory, positioning the product favorably within the consumer’s evoked set.
• Subjective Favorability (Unfavorable / Favorable): Represents a cognitive appraisal of suitability, desirability, and preference. Favorability indicates that the consumer views the product as aligning with their tastes or consumption goals, influenced by the advisor’s implied endorsement.
• Overall Quality Appraisal (Bad / Good): Captures an overarching qualitative assessment. In situations characterized by information asymmetry—such as fine arts, gourmet foods, or complex technical goods—consumers rely on the advisor’s endorsement as an informational surrogate for intrinsic quality, shifting their evaluation along the bad–good continuum.

Because the scale measures an immediate post-recommendation state, it avoids contamination from extraneous behavioral constraints (such as liquidity limits or accessibility issues) while reliably indexing the cognitive-affective endorsement that precedes behavioral compliance and purchase execution.

Theoretical Framework

The theoretical architecture underpinning the Post-Recommendation Product Attitude scale is grounded in three complementary frameworks within social psychology and consumer behavior: Signaling Theory, Source Credibility Theory, and the Elaboration Likelihood Model (ELM).

Signaling Theory

Originating in evolutionary biology and economics (Spence, 1973), signaling theory posits that in environments marked by information asymmetry, actors interpret observable behaviors or cues as signals of unobservable underlying qualities. When a consumer evaluates an unfamiliar product (e.g., an original piece of artwork or an artisanal food item), they face high diagnostic uncertainty. The recommendation provided by an advisor operates as a signal. D’Angelo and Valsesia (2023) demonstrated that when an advisor provides a combinatory recommendation (“You should try Product A with Product B”), the complexity of coordinating two items signals high competence and specialized domain expertise. This inferred expertise reassures the consumer, lowering perceived risk and enhancing the post-recommendation product attitude through a cognitive transfer of credibility.

Source Credibility and Advice Taking

The social psychological literature on source credibility (Hovland & Weiss, 1951) posits that messages originating from sources perceived as possessing high expertise and trustworthiness produce greater persuasion and attitude change. When an advisor’s recommendation strategy successfully establishes epistemic authority (Kruglanski et al., 2005), consumers are more inclined to accept the advice. In the judge-advisor system (JAS) paradigm, individuals systematically adjust their initial opinions toward an advisor’s suggestion to the degree that they perceive the advisor as competent. The post-recommendation attitude captures the successful outcome of this opinion revision process.

Dual-Process Models of Persuasion

According to the Elaboration Likelihood Model (Petty & Cacioppo, 1986), persuasion can occur via central or peripheral routes. In advisory interactions, the post-recommendation product attitude can reflect either thorough processing of the product attributes highlighted by the advisor (central route) or heuristic reliance on the advisor’s status, delivery style, or recommendation format as a peripheral cue. The bipolar semantic differential items utilized by D’Angelo and Valsesia capture the unified evaluative output regardless of whether the processing route was primarily deliberative or heuristic.

Validity

The Post-Recommendation Product Attitude scale demonstrates rigorous psychometric validity across multiple experimental investigations conducted by D’Angelo and Valsesia (2023), including field and laboratory studies involving diverse consumer goods.

Construct and Convergent Validity

Construct validity is substantiated by the scale’s alignment with established theoretical expectations. In Study 1 of D’Angelo and Valsesia (2023), conducted in the context of artwork evaluation, participants who received a combinatory recommendation from a gallery advisor exhibited significantly more favorable post-recommendation attitudes toward the focal painting compared to those receiving singular recommendations or no recommendation. This shift was statistically mediated by perceived advisor expertise, confirming that the scale accurately registers variations in persuasion driven by underlying theoretical mechanisms.

Furthermore, convergent validity is evidenced by strong correlations between the three-item composite and related downstream behavioral intentions. Across studies, post-recommendation attitude was positively correlated with:

• Purchase intentions for the focal product (typical r values ranging from .68 to .78, p < .001).
• Willingness to pay (WTP), wherein higher attitude scores predicted increased financial valuation of the recommended item.
• Recommendation acceptance rates, measured as actual choice behavior in consequential selection tasks.

Discriminant Validity

The scale demonstrates clear discriminant validity against conceptually related but distinct constructs. Specifically, empirical analyses revealed that:

• Advisor Warmth vs. Product Attitude: While combinatory recommendations systematically increased perceived advisor expertise and subsequently post-recommendation product attitude, they did not alter perceived advisor warmth (friendliness, sociability), demonstrating that the scale does not simply capture a diffuse “halo effect” of positive social sentiment.
• Pre-existing Preferences vs. Post-Recommendation Attitude: In baseline-controlled experimental designs, the scale reliably distinguished between general baseline category interest and the incremental evaluative gain directly attributable to the advisor’s recommendation.
• Perceived Product Price vs. Attitude: Post-recommendation attitude operated independently of perceived product expensiveness, confirming that favorable attitudes were not merely proxies for inferred luxury or price tags.

Reliability

The reliability of the Post-Recommendation Product Attitude scale is supported by internal consistency metrics across experimental testing conditions. Because the instrument uses three highly focused, universally recognized semantic evaluative markers, random measurement error is substantially minimized.

Internal Consistency

In the primary empirical validation studies reported by D’Angelo and Valsesia (2023), the internal consistency of the three items was consistently high:

• Study 1 (Art Recommendation Context): The three-item scale demonstrated high internal reliability with a Cronbach’s alpha of α = .93. Inter-item correlations among the three semantic differential items exceeded r = .80, indicating a uniform evaluative underlying dimension.
• Study 2 (Food & Beverage / Cookie Context): Replicating the scale across a sensory hedonic category (cookies paired with hot beverages), the scale achieved a Cronbach’s alpha of α = .94, with average inter-item correlations exceeding r = .82.
• Composite Reliability: Across supplemental studies reported in the research, Composite Reliability (CR) values regularly exceed .92, well above the conventional psychometric threshold of .70 recommended by Nunnally and Bernstein (1994).

Test-Retest and Measurement Stability

Because the scale is primarily designed as a state-based evaluative measure for immediate post-intervention deployment, standard long-term test-retest reliability is less applicable, as attitudes naturally update as consumption experiences occur. However, within experimental control conditions where no new product information was introduced, repeated immediate administration demonstrated high short-term stability (r > .85), confirming that the instrument yields stable measurement in the absence of exogenous advisory shocks.

Factor Analysis

The factor structure of the three-item semantic differential scale has been confirmed using both exploratory and confirmatory factor analytic approaches across consumer samples.

Exploratory Factor Analysis (EFA)

When the three items (Negative/Positive, Unfavorable/Favorable, Bad/Good) are subjected to exploratory factor analysis using principal axis factoring or maximum likelihood extraction with unrotated solutions:

• A single dominant factor emerges with an eigenvalue typically exceeding 2.55 (accounting for over 85% to 90% of the total item variance).
• No secondary factor achieves an eigenvalue above 0.30, strictly satisfying the Kaiser-Guttman criterion and scree plot examination for unidimensionality.
• Factor loadings for all three items are consistently high, ranging between .88 and .96, indicating that each pair reflects the core evaluative construct.

Confirmatory Factor Analysis (CFA)

In structural equation modeling (SEM) and CFA frameworks evaluating measurement models for post-recommendation consumer evaluations, the single-factor specification exhibits good statistical fit:

• Standardized Factor Loadings:
  • Negative / Positive: λ = .91 – .95
  • Unfavorable / Favorable: λ = .92 – .96
  • Bad / Good: λ = .89 – .94
• Model Fit Indices: For a just-identified (saturated) three-indicator single-factor model (df = 0), the model accounts for the observed covariance matrix. When embedded within broader structural models containing antecedent constructs (e.g., perceived expertise) and consequent variables (e.g., behavioral intent), fit indices regularly satisfy established criteria (Hu & Bentler, 1999): Comparative Fit Index (CFI) > .98, Tucker-Lewis Index (TLI) > .97, Root Mean Square Error of Approximation (RMSEA) < .05, and Standardized Root Mean Square Residual (SRMR) < .03.
• Average Variance Extracted (AVE): The AVE routinely exceeds .82, significantly above the .50 benchmark, demonstrating that the vast majority of variance captured by the indicators is shared common variance rather than error.

Instrument / Measurement Tool

The structure and administrative specifications of the Post-Recommendation Product Attitude scale are detailed below:

• Instrument Name: Post-Recommendation Product Attitude
• Authors: Jennifer K. D’Angelo and Francesca Valsesia (2023)
• Construct Assessed: Immediate evaluative valence, favorability, and quality perception of a product following an advisory recommendation
• Test Type: Self-report psychometric scale
• Number of Items: 3 bipolar adjective pairs
• Response Format: 7-point semantic differential scale (1 to 7)
• Administration Time: Less than 1 minute (approximately 20–30 seconds)
• Target Population: Consumers, experimental participants, retail patrons, and digital users receiving recommendations
• Scoring and Aggregation Rules:
  • Each item is scored on an integer scale from 1 (most negative adjective) to 7 (most positive adjective).
  • All three items are oriented in the same direction (negative pole = 1, positive pole = 7); therefore, no reverse scoring is required.
  • The individual scores of the three items are mathematically averaged to create a single composite post-recommendation attitude index: (text{Attitude} = frac{text{Item}_1 + text{Item}_2 + text{Item}_3}{3}).
  • Scores range from 1.00 to 7.00, with higher values reflecting more positive, favorable, and approving post-recommendation attitudes.

Permissions, Fee, and Test Year

• Year of Publication: 2023
• Original Publication: Journal of Marketing Research, Volume 60, Issue 1, pages 155–169
• Copyright Holder: American Marketing Association (AMA) / SAGE Publications
• Academic Permissions & Licensing: The scale items consist of standard semantic differential adjectives widely utilized within empirical consumer research. Under standard academic fair use principles, researchers may administer and utilize the scale for non-commercial educational, scholarly, and experimental research purposes without monetary fee, provided that appropriate bibliographic attribution is accorded to the original authors (D’Angelo & Valsesia, 2023). Commercial implementations, proprietary diagnostic use, or reprinting within commercial assessment batteries may require formal permission through the Copyright Clearance Center (CCC) or the American Marketing Association.

References

• D’Angelo, J. K., & Valsesia, F. (2023). You should try these together: Combinatory recommendations signal expertise and improve product attitudes. Journal of Marketing Research, 60(1), 155–169. https://doi.org/10.1177/00222437221110041
• Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Harcourt Brace Jovanovich College Publishers.
• Hovland, C. I., & Weiss, W. (1951). The influence of source credibility on communication effectiveness. Public Opinion Quarterly, 15(4), 635–650. https://doi.org/10.1086/266350
• 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
• Kruglanski, A. W., Raviv, A., Bar-Tal, D., Raviv, A., Sharvit, K., Ellis, S., Sadot, R., Pierro, A., & Mannetti, L. (2005). Says who? Epistemic authority and its role in social judgment. Advances in Experimental Social Psychology, 37, 345–392. https://doi.org/10.1016/S0065-2601(05)37006-7
• Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill.
• Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning. University of Illinois Press.
• Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances in Experimental Social Psychology, 19, 123–205. https://doi.org/10.1016/S0065-2601(08)60214-2
• Spence, M. (1973). Job market signaling. The Quarterly Journal of Economics, 87(3), 355–374. https://doi.org/10.2307/1882010

Items of the Scale