Voters' Involvement, Attitude, and Confidence in the Era of New Media

The study examines how determinants of social network site usage—need to belong, perceived ease of use, and perceived usefulness—relate to voters’ involvement with politics and, in turn, shape voters’ attitudes toward voting and confidence in their voting decisions. Motivated by the growing role of social media in election campaigns (e.g., Obama 2008; Trump 2016) and concerns such as microtargeting and misinformation (e.g., Cambridge Analytica; widespread online misinformation in Taiwan), the paper defines voters’ involvement with politics as the perceived importance and personal relevance of politics. Prior literature suggests involvement influences attitude and confidence. The research proposes that involvement mediates the effects of the social network site usage determinants on voting attitudes and confidence. Research questions: (1) Do determinants of social network site usage influence users’ involvement with politics? (2) What is the relationship between involvement with politics, attitude toward voting, and confidence in voting decisions in the context of social network sites? Hypotheses: H1: Social network site usage determinants positively influence voters’ involvement with politics. H2: Involvement is positively related to voting attitude. H3: Involvement is positively related to confidence in voting decisions. H4: Involvement mediates the effects of usage determinants on attitude toward voting. H5: Involvement mediates the effects of usage determinants on confidence in voting decisions.

The study builds on: (a) Need to belong theory (Baumeister & Leary, 1995) and its application to social media use (Seidman, 2013), positing that belongingness needs drive engagement with social network sites for communication, information, approval, and influence. (b) Technology Acceptance Model (Davis, 1989; operationalized by Kwon & Wen, 2010) indicating perceived ease of use and perceived usefulness predict social network site use. (c) Involvement literature defining involvement as a psychological state regarding importance and personal relevance (Hartwick & Barki, 1994; O’Cass & Fenech, 2003) and linking involvement to attitudes (Ajzen & Fishbein, 1977; Jackson et al., 1997) and to decision confidence (O’Cass & Pecotich, 2005; Zechman, 1979). Prior work shows that need to belong, perceived ease of use, and perceived usefulness correlate with social network site use, but their downstream influence on political involvement, voting attitudes, and decision confidence had not been directly tested in a mediated framework. The paper extends TAM and belongingness perspectives to voters’ mental processing in political contexts.

Design: Cross-sectional online survey with Likert five-point scales (1 = strongly disagree to 5 = strongly agree). Population and sampling: Facebook users of voting age (≥20) in Taiwan; recruitment via Facebook with an online survey platform. Data collection yielded 400 responses; after excluding ineligible/incomplete responses and follow-ups, 211 valid questionnaires remained. Nonresponse bias was assessed via early–late comparison (Armstrong & Overton, 1977); t-tests indicated no significant differences. Sample demographics were similar to Taiwan Facebook users (female ~55%, male ~45%, largest age group 18–34). Measures: • Need to belong: 5 items from Seidman (2013), α = 0.808. • Perceived ease of use: 5 items from Kwon & Wen (2010), α = 0.898. • Perceived usefulness: 5 items from Kwon & Wen (2010), α = 0.915. • Voters’ involvement with politics: 5 items adapted to politics (from prior scales), α = 0.934. • Voters’ attitude toward voting: 4 items (Pinkleton et al., 2002), α = 0.914. • Voters’ confidence in decisions: 3 items (O’Cass, 2002), α = 0.947. Controls: age, gender, education, occupation, monthly income (single-item each). Reliability: Factors with Cronbach’s α < 0.7 would be eliminated (none of the final scales were below 0.7). Multicollinearity: Variance inflation factors (VIFs) ranged 1.330–2.710 (mean 1.950), indicating no multicollinearity concerns. Analysis: Hierarchical regression models tested direct and mediated effects, controlling for demographics. Base models (1a, 2a, 3a) included controls only; extended models added predictors. Mediation was tested following Baron & Kenny (1986) with Sobel tests for indirect effects. Key model details: • Model 1b: Involvement regressed on need to belong, perceived ease of use, perceived usefulness; ΔR² = 0.447 over controls. • Models 2c and 3c: Attitude and confidence regressed on involvement. • Models 2b and 3b: Attitude and confidence regressed on determinants (without mediator). • Models 2d and 3d: Attitude and confidence regressed on determinants plus mediator to assess mediation. Additional fit changes with mediator: ΔR² = 0.172 (attitude), ΔR² = 0.101 (confidence). Significance thresholds: + p<0.1, * p<0.05, ** p<0.01, *** p<0.001.

• H1 supported: Social network site usage determinants positively predict voters’ involvement with politics (Model 1b). Coefficients: need to belong β = 0.342, p < 0.001; perceived ease of use β = 0.324, p < 0.001; perceived usefulness β = 0.153, p < 0.05; model explained an additional ΔR² = 0.447 over controls. • H2 supported: Involvement positively predicts voting attitude (Model 2c): β = 0.675, p < 0.001. • H3 supported: Involvement positively predicts confidence in voting decisions (Model 3c): β = 0.581, p < 0.001. • Mediation (H4, H5) supported: When involvement is included (Models 2d, 3d), direct effects of determinants on outcomes reduce or become nonsignificant, indicating mediation. - Attitude (Model 2d): direct effects became nonsignificant (need to belong β = 0.069; perceived ease of use β = 0.072; perceived usefulness β = 0.052), with improved model fit (ΔR² = 0.172). - Confidence (Model 3d): direct effects diminished (need to belong β = 0.060; perceived ease of use β = 0.191, p < 0.01; perceived usefulness β = 0.007), with improved fit (ΔR² = 0.101). • Sobel tests confirmed significant indirect effects via involvement: - Need to belong: z = 5.03 (attitude), p < 0.001; z = 4.79 (confidence), p < 0.001. - Perceived ease of use: z = 4.69 (attitude), p < 0.01; z = 4.50 (confidence), p < 0.001. - Perceived usefulness: z = 2.24 (attitude), p < 0.05; z = 2.22 (confidence), p < 0.05. • Control models with demographics alone showed low explanatory power (e.g., R² ≈ 0.020–0.071) except gender showing some direct associations in base models; full models substantially improved explanatory power.

The findings address the research questions by demonstrating that the determinants of social network site usage—need to belong, perceived ease of use, and perceived usefulness—enhance voters’ political involvement, which subsequently increases favorable attitudes toward voting and confidence in voting decisions. This underscores the central role of involvement as a psychological mechanism linking social media use determinants to political outcomes. The results extend TAM and need-to-belong theory to a political context, showing that platform-related perceptions and social motivations have downstream effects on voters’ internal evaluations and perceived decision competence. Practically, political campaigners should not assume that mere social media access shapes attitudes and confidence; instead, strategies should cultivate users’ sense of belonging and ensure ease of use to raise political involvement, thereby improving attitudes toward voting and decision confidence. The study provides empirical evidence using real social network site users in Taiwan, highlighting the relevance of social media in contemporary elections, especially in environments with high platform penetration and exposure to misinformation.

The study proposes and tests a mediated framework linking social network site usage determinants (need to belong, perceived ease of use, perceived usefulness) to voters’ political involvement, which in turn drives voting attitudes and decision confidence. Using survey data from 211 Facebook users in Taiwan and regression-based mediation analyses, all hypotheses were supported: determinants positively influenced involvement; involvement positively affected attitudes and confidence; and involvement mediated the effects of determinants on both outcomes. Contributions include an integrated conceptual model bridging TAM and belongingness with political psychology, and practical insights for campaign strategy design centered on enhancing belonging and ease of use to raise involvement. Future research could validate and extend the model across different platforms and countries, employ probability sampling, and examine moderators such as personality traits and information content characteristics.

• Generalizability: The sample includes only Facebook users in Taiwan; findings may not generalize to other platforms or populations. • Sampling: Voluntary participation may introduce self-selection bias; probability sampling is recommended for future studies. • Scope of variables: Other factors (e.g., personality, content type/quality) may influence attitudes and confidence; future work should test moderating effects and additional determinants. • Cross-sectional design: Causal inferences are limited; longitudinal or experimental designs could strengthen causal claims.