Skip to content
Detection methods credibilityonline behaviorself efficacy

Social media self-efficacy and information evaluation online

Computers in Human Behavior · 2014

In brief

Hocevar et al. introduce social media self-efficacy—perceived ability to perform desired functions in social media—and examine its relationship to online credibility judgment. A survey of 3,568 US adults found that higher social media self-efficacy predicts perceiving social media information as more trustworthy, relying more on others' opinions during evaluation, and greater use of social media to verify information.

Summary figure for Social media self-efficacy and information evaluation online

Contributions

  • Social media self-efficacy construct: Proposes that self-efficacy, operationalized as perceived social media skill, confidence in finding information online, and content production/consumption frequency, is distinct from general internet self-efficacy.
  • Self-efficacy as credibility predictor: Demonstrates that higher social media self-efficacy positively predicts perceived trustworthiness of social media information relative to offline sources ($r = .24$, $p < .001$).
  • Domain moderation effect: Shows the relationship is stronger for product information ($\beta = .76$) than health ($\beta = .48$) or news ($\beta = .43$), possibly due to greater potential harm from health misinformation.
  • Social evaluation strategies: Higher self-efficacy users more frequently consult others' opinions ($r = .30$, $p < .001$) and use social media specifically to verify information ($r = .50$, $p < .001$).

Method

Sample: Representative survey of 3,568 US adult internet users via Knowledge Networks' probability-based panel. Survey developed through iterative pre-testing to ensure comprehension of social media definitions and items.

Measurement:

  • Social media self-efficacy: Composite of perceived skill (11-point scale), confidence finding information (11-point scale), content production (8 items, 5-point scale, $\alpha = .89$), and content consumption (5 items, 5-point scale, $\alpha = .81$). Normalized across components and averaged ($\alpha = .78$).
  • Perceived trustworthiness: Single item per domain (product, health, news) on 7-point scale comparing social media to offline sources.
  • Social credibility assessment: Four items, 5-point scale ($\alpha = .83$), measuring reliance on others' opinions during credibility judgment.
  • Use of social media for verification: Averaged responses across five social media sources (blogs, Wikipedia, ratings sites, etc.) on 5-point likelihood scale ($\alpha = .90$).

Analysis: General linear model (GLM) for moderation hypotheses; simple slope analysis to probe interactions.

Results

H1 (supported): Self-efficacy positively predicts trustworthiness perceptions ($r = .24$, $p < .001$). Aligns with prior work showing familiarity and prior experience influence trust judgments.

H2 (supported): Domain moderates the relationship. Product information shows the steepest slope; health and news show weaker relationships, consistent with barriers-to-trust theory (greater potential harm = higher distrust of unfamiliar sources).

H3 (supported): Self-efficacy predicts social credibility assessment frequency ($r = .30$, $p < .001$). Effect size for self-efficacy ($\eta_p^2 = .10$) substantially larger than domain effect ($\eta_p^2 = .02$), suggesting self-efficacy is the dominant predictor regardless of information type.

RQ1 (not significant): No significant interaction between self-efficacy and domain in predicting social credibility assessment, $F(5,3533) = 1.90$, n.s.

RQ2 (supported): Domain moderates the relationship between self-efficacy and social media verification use, $F(5,3487) = 3.37$, $p < .01$, $\eta_p^2 = .01$. At low self-efficacy, domain-specific information (news, entertainment, health, product) is less likely to be verified via social media than general information. Effect persists at high self-efficacy but is attenuated.

Notes

Strengths:

  • Large, representative sample with rigorous survey design and pre-testing.
  • Introduces a psychologically grounded construct with strong empirical grounding in Bandura's theory.
  • Examines both outcomes (trustworthiness perception) and processes (social credibility assessment, verification behavior), not just attitudes.
  • Domain-moderated analysis reveals important nuance: self-efficacy's effect is not uniform across information types.
  • Addresses practical risk: self-efficacious users may be particularly vulnerable to trusting unverified social media content, especially in health domains where accuracy carries real consequences.

Limitations and open questions:

  • Cross-sectional design; cannot establish causality (does experience build self-efficacy, or does existing self-efficacy lead to more social media use?).
  • Trustworthiness measured as perceived trustworthiness relative to offline sources, not actual accuracy of information trusted. Users higher in self-efficacy may trust misinformation.
  • Social media self-efficacy measured as aggregated across all platforms; venue-specific efficacies (blogs vs. wikis vs. Twitter) might show different relationships.
  • Sample is US-based internet users in 2014; findings may not generalize to other cultures or current mobile-first social media landscape.
  • Authors note potential confound: similarity with other social media users (perceived shared identity/interests) may predict both self-efficacy and trust, but was not directly measured.

Implications for the field:

  • Extends self-efficacy theory beyond traditional performance domains into social judgment and credibility assessment.
  • Suggests information literacy interventions might need to target not just skills but also the overconfidence that accompanies skill, particularly in health/news contexts where stakes are high.
  • Relevant to understanding echo chambers: users confident in their ability to evaluate social media may be less likely to seek expert corroboration.

How it connects

Similar papers

Summarized from the full text. Added 2026-11-20 from a submission by reza. Every statement on this page traces back to the source.