Privacy of Information (Withholding Personal Information) (POI)

Abstract

The Privacy of Information (Withholding Personal Information) (POI) scale is a concise, highly focused psychometric instrument developed by May O. Lwin, Jochen Wirtz, and Jerome D. Williams in 2007 to quantify defensive consumer behavior in digital environments. Specifically, the instrument assesses an individual’s behavioral intention to deliberately withhold personal data, abandon registration processes, or completely terminate interactions with a specific website in response to perceived threats to information privacy. Grounded in the Power-Responsibility Equilibrium framework, the social contract theory, and consumer protection paradigms, the scale captures non-cooperative behavioral avoidance as an active privacy management strategy. The POI is a unidimensional measure comprising three items administered on a 7-point Likert-type response format, anchored from strongly disagree (or very unlikely) to strongly agree (or very likely). Extensive empirical validation across consumer research and digital marketing studies demonstrates superior psychometric properties: Cronbach’s alpha values consistently exceed 0.85 (reported at α = 0.89 in the seminal validation study), composite reliability is established above 0.88, and average variance extracted (AVE) exceeds 0.70. Confirmatory factor analysis demonstrates distinct discriminant validity against related reactive constructs such as information fabrication, data falsification, regulatory complaints, and protective software adoption. The instrument serves as a critical diagnostic and evaluative tool for researchers, organizational psychologists, e-commerce strategists, and human-computer interaction (HCI) scholars seeking to model how online trust deficits, corporate data collection policies, and privacy notices drive behavioral disengagement.

Keywords

Privacy of Information, Withholding Personal Information, Online Privacy, Consumer Privacy Protection, Defensive Privacy Behavior, Information Disclosure, Digital Disengagement, Power-Responsibility Equilibrium, E-commerce Trust, Psychometrics.

Authors

The scale was developed and psychometrically validated by a multidisciplinary team of scholars specializing in marketing, consumer psychology, public health communication, and behavioral economics:

• May O. Lwin, Ph.D. — Professor of Communication and Information, Wee Kim Wee School of Communication and Information, Nanyang Technological University (NTU), Singapore. Expert in strategic communication, digital health interventions, and cyber-risk perceptions.
• Jochen Wirtz, Ph.D. — Professor of Marketing and Vice Dean of Graduate Studies, NUS Business School, National University of Singapore (NUS), Singapore. Internationally recognized authority on services marketing, customer relationship management, and technology adoption.
• Jerome D. Williams, Ph.D. (1947–2021) — Distinguished Professor and Prudential Chair in Business, Rutgers Business School, Rutgers University, Newark and New Brunswick, USA. Renowned scholar in consumer behavior, multicultural marketing, and marketplace public policy.

Purpose

The primary purpose of the Privacy of Information (Withholding Personal Information) scale is to measure an Internet user’s self-reported intention or propensity to deliberately refrain from revealing personal identifiers, demographic details, or sensitive behavioral data when interacting with a digital interface or commercial platform. As online commerce and digital services expanded exponentially during the early 2000s, organizations increasingly collected vast quantities of granular personal data, often without reciprocal protections, transparent consent mechanisms, or verifiable safeguards. Prior research largely focused on general cognitive attitudes, such as abstract privacy concerns or generalized internet cynicism. However, general concerns frequently failed to predict actual behavioral manifestations—a paradox widely documented in behavioral economics as the “privacy paradox.”

To overcome this empirical disconnect, Lwin, Wirtz, and Williams (2007) introduced behavioral response metrics that operationalize concrete, action-oriented protective maneuvers undertaken by consumers when a website’s perceived risk outweighs its perceived benefits. The POI specifically isolates the construct of passive resistance or protective withholding. Unlike active deceptive strategies (e.g., providing fabricated names, disposable email addresses, or falsified postal codes), withholding represents an overt termination of information exchange or outright refusal to transact. This behavioral response directly impacts digital business models, lead generation, user conversion rates, and online service personalization algorithms.

In academic and clinical research contexts, the POI scale is utilized to evaluate how user interface designs, cookie notices, explicit opt-in/opt-out mechanisms, and third-party security seals modulate user apprehension. In organizational research, it functions as a sensitive criterion variable in structural equation modeling (SEM) to assess the consequences of perceived organizational justice, corporate reputation loss, and data breach disclosures. From a clinical and psychological perspective, the instrument facilitates investigations into digital paranoia, general risk aversion, locus of control, and personal boundary regulation within increasingly ubiquitous computing ecosystems.

Psychological Construct

The construct captured by the POI scale is situated within the broader domain of consumer self-protection and boundary regulation mechanisms. In environmental and developmental psychology, boundary regulation refers to an individual’s ongoing dynamic process of managing personal accessibility to others (Altman, 1975). When applied to human-computer interaction and digital exchange, information privacy represents the ability of individuals to control the terms, conditions, and extent to which their personal information is acquired, stored, shared, or exploited by external entities.

The POI construct specifically represents information withholding as an avoidance coping mechanism. Coping theory categorizes behavioral adaptations into problem-focused coping (direct action to alter the stressful environment) and emotion-focused or avoidance coping (withdrawing from the source of stress). Withholding personal information reflects a direct, problem-focused strategy designed to eliminate personal vulnerability at the transaction boundary. When an individual confronts an intrusive online data capture form—such as mandatory requirements for phone numbers, home addresses, government identification numbers, or lifestyle habits—they evaluate the transaction through cognitive appraisal processes:

• Perceived Intrusion: The degree to which data requests violate contextual expectations of relevance and necessity for the requested service.
• Calculated Exposure Risk: The perceived probability of unauthorized secondary data use, spam generation, identity theft, or tracking by third-party data brokers.
• Decision to Withhold: The conscious behavioral decision to exit the website, abandon an active transaction basket, or deliberately leave mandatory and optional data fields empty, accepting the opportunity cost of not receiving the website’s content or service.

Importantly, the construct of withholding is conceptually and empirically distinct from related privacy management behaviors. It must be differentiated from information fabrication (actively feeding false data into the system to poison data profiles or gain unauthorized access), protective software adoption (deploying ad-blockers, tracking prevention extensions, or virtual private networks), and complaint behavior (filing formal grievances with corporate data protection officers or state regulatory commissions). Withholding personal information is characterized by transactional cessation; the individual asserts autonomy through complete non-participation, rendering data extraction technically impossible at that specific digital touchpoint.

Theoretical Framework

The POI scale is theoretically anchored in the Power-Responsibility Equilibrium (PRE) model, synthesized with social contract theory and protection motivation theory (PMT). The seminal work of Lwin, Wirtz, and Williams (2007) adapts the classical sociological concept of power-responsibility balance to commercial digital ecosystems.

Under the PRE paradigm, parties in an exchange relationship operate within an implicit equilibrium where power must be commensurate with institutional responsibility. When an online enterprise exercises significant power—through structural information asymmetry, proprietary algorithmic control, and mandatory data intake gates—consumers expect the enterprise to exhibit a corresponding level of fiduciary responsibility. This responsibility includes deploying transparent disclosure policies, state-of-the-art encryption, explicit consent gathering, and strict adherence to purpose limitation. When organizations fail to exhibit this responsibility, consumers perceive a severe structural imbalance. In response to perceived power exploitation, consumers deploy reactive coping mechanisms to restore parity or mitigate personal risk, among which information withholding is the primary defensive response.

Complementing PRE, social contract theory posits that economic transactions rely upon mutually understood normative rules. When a consumer visits a website, an implied social contract governs the interaction: the user provides attention, engagement, or personal details in return for relevant content, entertainment, or commercial utility. When a platform requests data perceived as disproportionate, intrusive, or dangerous to personal safety without offering equitable value or protection, the social contract is breached. The psychological consequence of this perceived breach is psychological reactance and immediate relational severance, manifesting behaviorally as withholding.

Finally, Protection Motivation Theory explains the cognitive mechanisms initiating the withholding behavior. PMT conceptualizes response through threat appraisal (perceived severity of identity theft, commercial profiling, or financial fraud, alongside personal vulnerability) and coping appraisal (response efficacy of withholding and self-efficacy to execute the avoidance). Because withholding personal information has near-total response efficacy (an individual cannot be compromised by data they never provide) and requires minimal technological sophistication compared to setting up complex proxy networks or encrypted virtual machines, it emerges as the most immediate, accessible, and pervasive behavioral coping strategy across universal consumer demographics.

Validity

Empirical evaluations of the POI scale demonstrate robust evidence of construct, convergent, discriminant, and predictive validity across diverse samples and experimental conditions.

Construct and Convergent Validity

In the original validation study by Lwin, Wirtz, and Williams (2007), which involved a multi-factorial experimental design utilizing diverse online scenarios, the POI items loaded heavily and unambiguously onto their designated latent factor. Factor loadings ranged from 0.82 to 0.91, exceeding the conventional psychometric threshold of 0.70. The calculated average variance extracted (AVE) for the withholding factor was 0.74, well above the recommended benchmark of 0.50 established by Fornell and Larcker (1981). This confirms that more than 70% of the variance captured by the indicators is direct construct variance rather than measurement error.

Discriminant Validity

Discriminant validity was established via rigorous comparison against related digital privacy behavior constructs measured simultaneously in the research design: Fabrication of Personal Information (e.g., providing fake demographics or erroneous email addresses) and Protection via External Means (e.g., lodging complaints or engaging privacy-enhancing technologies). Using the Fornell-Larcker criterion, the square root of the AVE for Withholding (√0.74 = 0.86) was substantially greater than its bivariate correlations with fabrication (r = 0.42) and complaint intentions (r = 0.38). Further structural equation modeling demonstrated that constraining the correlation between withholding and fabrication to unity resulted in a statistically significant degradation of chi-square goodness-of-fit (Δχ² > 85.0, p < 0.001), corroborating that withholding is empirically distinct from active data falsification.

Nomological and Predictive Validity

Nomological validity is verified by the scale’s predictable relationships with upstream antecedents and downstream outcomes:

• Antecedent Predictors: As predicted by PRE theory, high perceived commercial privacy concern (β = 0.48, p < 0.001) and high perceived procedural unfairness (β = 0.39, p < 0.001) positively predict withholding behavior. Conversely, the presence of explicit opt-in consent mechanisms and trust seals exhibits a strong negative relationship with withholding (β = -0.34, p < 0.01).
• Predictive Efficacy: Experimental research validates that high scores on the POI scale directly correspond to real-world behavioral outcomes, including digital shopping cart abandonment, lower registration conversion rates, and reduced time spent on data-intrusive landing pages.

Reliability

The POI scale exhibits outstanding internal consistency across various experimental samples, survey panels, and cultural environments. The original investigation reported the following statistical parameters:

• Internal Consistency: The scale achieved a Cronbach’s alpha of α = 0.89 in the primary study sample (N = 432), demonstrating high internal homogeneity among the three items.
• Composite Reliability: The composite reliability (CR) coefficient was calculated at CR = 0.89, comfortably exceeding the standard threshold of 0.70 advocated for behavioral research.
• Item-Total Correlations: Corrected item-to-total correlations for each of the three scale items were robust, ranging from 0.76 to 0.83, confirming that every item contributes substantially to the underlying construct without redundancy.
• Cross-Sample Stability: Subsequent cross-national studies applying the scale in North American, European, and Asian online shopping environments have consistently documented alpha coefficients ranging between 0.84 and 0.93. Test-retest reliability across a two-week interval in consumer panel testing yielded an intraclass correlation coefficient (ICC) of 0.81 (p < 0.001), indicating strong temporal stability of the scale when platform conditions remain unchanged.

Factor Analysis

The latent structure of the POI scale has been thoroughly confirmed through both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA).

Exploratory Factor Analysis (EFA)

During preliminary scale purification, principal axis factoring with promax (oblique) rotation was conducted on a combined pool of consumer defense items. The analysis revealed a clean three-factor solution representing Withholding, Fabrication, and Direct Protection/Complaining. The three items designated for Withholding Personal Information loaded uniquely onto a single factor with an eigenvalue of 4.12, accounting for over 68% of the explained variance among the defense response indicators. Factor loadings were exceptionally clean:

• Item 1 (Refrain from registering): Loading = 0.88
• Item 2 (Decide not to use the site): Loading = 0.85
• Item 3 (Refuse to provide personal details): Loading = 0.90

Cross-loadings onto other factors were negligible (all cross-loadings < 0.18), demonstrating unambiguous factor purity.

Confirmatory Factor Analysis (CFA)

Subsequent confirmatory factor analysis performed using maximum likelihood estimation in AMOS and LISREL substantiated the unidimensionality of the three-item withholding construct. The measurement model demonstrated excellent fit indices:

• Goodness-of-Fit Index (GFI): 0.98
• Comparative Fit Index (CFI): 0.99
• Tucker-Lewis Index (TLI): 0.98
• Standardized Root Mean Square Residual (SRMR): 0.021
• Root Mean Square Error of Approximation (RMSEA): 0.038 (90% CI: [0.000, 0.062])
• Model Chi-Square / Degrees of Freedom: χ²/df = 1.45 (p > 0.05, indicating non-significant divergence from empirical data)

All standardized path coefficients from the latent construct to the observed indicators were statistically significant at p < 0.001, providing empirical justification for retaining all three items in a single, unweighted composite score.

Instrument / Measurement Tool

The POI instrument is structured for rapid, low-burden administration in both laboratory experiments and field surveys. Its characteristics are summarized below:

• Instrument Name: Privacy of Information (Withholding Personal Information) Scale (POI)
• Construct Assessed: Consumer behavioral intention to withhold personal data and discontinue website interaction due to perceived privacy risks
• Number of Items: 3 items
• Administration Format: Self-administered online questionnaire, digital survey, or paper-and-pencil scale
• Administration Time: Less than 2 minutes
• Response Options: 7-point Likert-type scale typically anchored from 1 (“Strongly Disagree” or “Very Unlikely”) to 7 (“Strongly Agree” or “Very Likely”). Some researchers employ a 5-point variant without significant loss of psychometric sensitivity.
• Scoring Protocol: All three items are positively worded; therefore, no reverse scoring is required. The overall withholding score is computed either as the mathematical mean of the three items (ranging from 1.00 to 7.00) or as a summated total (ranging from 3 to 21).
• Score Interpretation:
  • Low Withholding Propensity (Mean 1.00 – 2.99): The user feels secure, trusts the platform’s data management practices, perceives the value exchange as favorable, and is willing to disclose required personal data.
  • Moderate Withholding Propensity (Mean 3.00 – 4.99): Ambivalence regarding data security; user may selectively provide basic information while hesitating to complete full onboarding.
  • High Withholding Propensity (Mean 5.00 – 7.00): Severe privacy apprehension, trust failure, or perceived procedural injustice; the user actively rejects data collection, leading to site abandonment or registration refusal.

Permissions & Fee and Test Year

The Privacy of Information (Withholding Personal Information) scale was formally introduced in 2007 in the Journal of the Academy of Marketing Science. The scale was developed within an academic research context supported by institutional university grants.

Copyright and Usage Permissions: The original publication is copyrighted by Springer Science+Business Media and the Academy of Marketing Science. In accordance with conventional academic fair-use guidelines, the three items may be utilized, adapted, and cited for non-commercial academic research, pedagogical purposes, and scientific inquiry without explicit written permission, provided that appropriate scholarly attribution is accorded to Lwin, Wirtz, and Williams (2007). Commercial organizations, market research firms, or corporate entities integrating the scale into proprietary commercial testing software or fee-based user research platforms should verify compliance with publisher licensing policies and obtain appropriate commercial permissions through the Copyright Clearance Center (CCC) or Springer Nature.

References

• Altman, I. (1975). The Environment and Social Behavior: Privacy, Personal Space, Territory, Crowding. Monterey, CA: Brooks/Cole Publishing Company.
• 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
• Lwin, M., Wirtz, J., & Williams, J. D. (2007). Consumer online privacy concerns and responses: A power–responsibility equilibrium perspective. Journal of the Academy of Marketing Science, 35(4), 572–585. https://doi.org/10.1007/s11747-006-0003-3
• Milne, G. R., Rohm, A. J., & Bahl, S. (2004). Consumers’ protection of online privacy and identity. Journal of Consumer Affairs, 38(2), 217–232. https://doi.org/10.1111/j.1745-6606.2004.tb00865.x
• Smith, H. J., Dinev, T., & Xu, H. (2011). Information privacy research: An interdisciplinary review. MIS Quarterly, 35(4), 989–1015. https://doi.org/10.2307/41409970
• Wirtz, J., & Lwin, M. O. (2009). Regulatory focus, trust, guarantees, and online information privacy. Journal of Business Research, 62(2), 190–198. https://doi.org/10.1016/j.jbusres.2008.01.027

Items of the Scale