Taste for Privacy: How Context, Identity, and Lived-Experience Shape Information Sharing Preferences

Abstract: Privacy preferences are not fixed individual traits, they depend on context and lived experiences. In this study, we analyze 2,912 survey responses from 782 college students collected over seven survey periods during 2023 and 2024. We ask about their usage of social media, the security settings of their accounts, and measure their comfort in sharing personally identifiable information (PII) across 17 different institutional contexts. Compared to past research, we observe a large shift towards private accounts, going from 1/3rd private in 2007 to 2/3rds in 2024, and find that participants' discomfort sharing PII with social media platforms strongly predicts their privacy settings. Beyond social media, we identify a stable ranking of institutional trust, though some institutions, like the police, show high variability reflecting divergent lived experiences. Traditionally marginalized groups and participants having faced adverse childhood experiences show more discomfort with institutions of power, especially in areas where they face greater vulnerability. We argue for context-adaptive privacy settings that recognize institutional relationships and demographic vulnerabilities, moving beyond one-size-fits-all consent frameworks toward contextually appropriate data governance.

• The paper presents a large-scale survey study of over 2,900 responses to reveal how contextual factors, demographic identity, and adverse childhood experiences influence privacy settings.
• It finds that increased discomfort with sharing PII, particularly on social media, is associated with a 37% higher likelihood of selecting private profiles.
• The study’s regression and dose-response analyses underscore the need for context-adaptive privacy controls in interface design and policymaking.

Contextual Privacy Preferences: Empirical Patterns and Implications

Introduction

The paper "Taste for Privacy: How Context, Identity, and Lived-Experience Shape Information Sharing Preferences" (2604.22025) engages in empirical analysis to characterize privacy preferences among college students, emphasizing the dynamic interplay between context, demographic identity, and lived experience. Diverging from trait-based privacy frameworks, the authors provide systematic evidence for contextual integrity by examining over 2,900 survey responses collected from 782 participants during 2023-2024. The study interrogates the relationship between privacy settings, institutional trust, demographic variation, and adverse childhood experiences (ACEs), and advocates for context-adaptive privacy governance.

Methodological Framework

The research leverages a multi-stage survey methodology embedded within the Lived Experiences Measured Using Rings Study (LEMURS). The participant pool consists of first-year undergraduate students at a Northeastern US institution. Data are drawn from seven survey rounds, capturing privacy behaviors (social media profile visibility, platform usage, frequency), as well as participants' comfort levels in sharing personally identifiable information (PII) across 17 institutional contexts via a 7-point Likert scale.

Analytical strategies include ordinal logistic regression to connect privacy settings and comfort levels, bootstrap-based difference-in-mean and rank analyses for demographic comparisons, and dose-response modeling of ACEs relative to institutional discomfort. Demographics were coded with one-hot encoding and categories were aggregated for statistical robustness.

Empirical Findings

Shift in Social Media Privacy

There is a pronounced shift in privacy behavior: the proportion of students with private social media profiles has doubled since 2007, from 33.2% to 62%, and an additional 22% report mixed settings. Public profiles now represent only 16%. Comfort with sharing PII with social media platforms robustly predicts usage of private settings. Each unit increase in discomfort on the Likert scale is associated with a 37% increment in the odds of choosing more private settings (OR=1.37, $p<0.001$). Even after controlling for platform usage and number of platforms, discomfort with social media remains the dominant predictor of privacy settings (OR=1.34, $p<0.001$).

Usage frequency and platform diversity inversely correlate with privacy: participants who use more platforms or use them more often are statistically less likely to implement private profiles (OR=0.82 for platform count, $p<0.001$; OR=0.90 for usage frequency, $p<0.02$).

Institutional Trust Hierarchy

Analysis of institutional trust reveals a stable hierarchy. Friends, medical professionals, and relatives consistently rank as the most trustworthy contexts for sharing PII, with mean Likert scores (reflecting lowest discomfort) of 2.23–2.65. Strangers and unfamiliar companies occupy the least trusted positions (mean Likert 5.87 and 5.52). Social media platforms are positioned between weak social ties and unfamiliar companies. This suggests reluctant participation and reinforces the inadequacy of undifferentiated privacy models.

Concentric trust gradients emerge: as social distance from the participant increases, discomfort grows. Institutions are grouped into high-trust (close personal), mid-tier (researchers, schools, employers, financial, nonprofits), interchangeable (co-workers, government, neighbors, police, customer companies, social media), and high-discomfort (strangers, non-customer companies).

Demographic and Lived-Experience Effects

Demographic analysis identifies patterned variation in institutional discomfort. Women report significantly greater discomfort with acquaintances, co-workers, neighbors, medical professionals, and relatives. Non-binary participants display heightened discomfort across both close and socially distant institutions, particularly police—this effect size is the largest observed across demographic axes.

People of color are more uncomfortable sharing PII with police, government, and strangers; first-generation students are more uncomfortable with financial institutions, schools, and researchers. Homosexual participants display greater discomfort with medical professionals and employers.

A dose-response relationship between ACE count and institutional discomfort is evident. Each additional ACE predicts increased discomfort with relatives, financial institutions, police, medical professionals, employers, government, and neighbors. The effect is concentrated in contexts characterized by power asymmetries and vulnerability, signifying that institutional failure or exposure to harm in childhood has durable effects on trust.

Implications for Theory and Data Governance

The empirical evidence robustly undermines privacy-as-trait models and universal consent frameworks. Instead, privacy preferences are contextually structured, shaped by institutional proximity, demographic identity, and lived experiences—consistent with contextual integrity theory. The results indicate the necessity for privacy controls that adapt to context and user vulnerability, in line with recent proposals for context-aware privacy interfaces and regulatory standards [6263765, 10.1145/3613904.3642500, shaffer2021applying, alamari2025adaptive].

Practically, platforms should deploy privacy settings modulated by relationship type and institutional trust hierarchy. Context-adaptive defaults and controls would enable users to articulate differentiated preferences, particularly salient for marginalized or traumatized populations. These findings also have ramifications for policy: consent should be granular, acknowledging heterogeneous discomfort across contexts, and interface design must increase transparency regarding institutional access to PII.

Theoretical implications include reconceptualizing privacy as a relational process that is calibrated across identity axes and experienced vulnerabilities. The cross-institutional consistency of trust hierarchies may inform computational modeling of privacy norms and automated policy generation. The study’s demonstration of digital resignation validates models such as Draper’s “corporate resignation,” implying that rational optimization of privacy is obstructed by social necessity and platform affordances.

Limitations and Future Directions

The sample is limited to college students at a single institution and may self-select for lower privacy tastes; generalizability to broader populations is constrained. Statistical grouping across demographic categories was necessary to ensure adequate power. Future work should leverage representative sampling, longitudinal designs, and mixed-methods interviews to deepen understanding of privacy decision-making and context-specific trust formation.

Further research is merited to dissect data-type preferences by institution, analyze networked privacy flows where individual choices are not independent, and develop algorithmic frameworks for adaptive privacy governance that operationalize measured trust hierarchies.

Conclusion

The paper provides rigorous empirical validation for contextual integrity in privacy preferences, demonstrating structured variation by institution, identity, and lived experience. The findings underscore the inadequacy of monolithic notice-and-consent mechanisms and motivate the design of context-adaptive privacy governance. Theoretical and practical advances in privacy must acknowledge relational dynamics, demographic vulnerabilities, and patterned distrust stemming from adverse experiences. These results offer actionable insights for interface designers, policymakers, and researchers seeking principled approaches to privacy in digitally mediated environments.