The research set out to address two gaps. Western theories of technological neutrality, which hold that digital tools carry no inherent political bias, and theories of algorithmic polarization, which argue that algorithmic curation traps users in filter bubbles and erodes political cohesion, were both developed in the context of Western multiparty systems and commercialized media. Hou argues that neither framework fully accounts for China’s state-led digital ecosystem, in which the government actively deploys AI-driven platforms to disseminate policy narratives. Meanwhile, much existing Chinese research on youth political identity had treated the construct as homogeneous, combining its dimensions into a single composite score without testing whether variables actually exert equal effects across them.

To break with that assumption, the study expanded a classic six-dimensional framework of Chinese citizens’ political identity into eight dimensions, adding emotional identity and institutional identity. The eight dimensions were then grouped into three clusters: an Emotional–Traditional Cluster comprising identity, emotional identity, and cultural identity, which the author characterizes as endogenous and rooted in long-standing emotional bonds; an Institutional–Performance Cluster comprising institutional, systemic, policy, and developmental identity, which is described as exogenous and driven by rational evaluations of governance outcomes; and Party Identity as a single, separate dimension. The questionnaire was administered online in a standardized, anonymous format, and after screening out invalid responses, 4,094 of 4,182 collected questionnaires were retained, a valid response rate of 97.9 percent.

Measurement relied on a mix of multi-item scales and single-item indicators, all on 5-point Likert scales. The independent variable, perceptions of digital and intelligent technologies, was measured with a self-developed seven-item scale covering technology adoption, development trends, security and ethics, impact on daily life, social impact, algorithmic risks, and literacy enhancement; it showed good reliability with a Cronbach’s alpha of 0.889. Five parallel mediators were tested: political participation, political communication, perception of rights, perception of interests, and political satisfaction. Only gender and place of origin were included as control variables in the regression and mediation models.

The analysis used a bootstrap multiple parallel mediation model with bias correction, drawing 5,000 repeated samples to estimate indirect effects at 95 percent confidence intervals, with robust standard errors to address heteroscedasticity. Confirmatory factor analysis supported the eight-dimensional structure, with average variance extracted above 0.5 and composite reliability above 0.7 for the key latent variables. The authors note one fit index, RMSEA of 0.083, slightly exceeded the conventional 0.08 threshold, but argue that chi-square-based indices are prone to rejecting models in very large samples and that robust indices such as CFI of 0.925 and SRMR of 0.054 support the measurement model.

The results revealed a clear hierarchy of transmission pathways. Political participation emerged as a significant positive mediator across all eight identity dimensions, with standardized coefficients ranging from 0.014 to 0.139, the strongest effects appearing for institutional identity at 0.139 and policy identity at 0.138. The correlation between digital technology literacy and political participation was 0.61, suggesting that students with higher digital literacy are more willing to engage in online political practices such as policy advocacy and digital deliberation, which in turn strengthens their sense of political efficacy and recognition of institutions and development.

Political communication, by contrast, showed marked stratified heterogeneity. Its mediating effect was not statistically significant for the Emotional–Traditional Cluster or for party identity, but was significantly positive within the Institutional–Performance Cluster, with coefficients between 0.030 and 0.090. Hou interprets this pattern through psychological reactance and social identity theory: one-way, push-style official messaging may be perceived as a threat to autonomy, triggering defensive reactions that block the internalization of emotionally rooted identities, whereas standardized government information about governance outcomes can still resonate with identities grounded in rational performance evaluation.

Two perceptual mediators behaved consistently across all dimensions. Perception of rights mediated positively everywhere, with coefficients from 0.030 to 0.096, indicating that students who feel their digital political rights are protected report stronger belonging and better evaluations of the political system. Perception of interests was the only consistently negative pathway, with coefficients from −0.017 to −0.040, producing what the author calls a masking effect. This variable had the lowest mean in the dataset at 2.63 and a large standard deviation of 1.505, and the study links it to algorithmic information environments that emphasize narrow personal interests, potentially leading students to exaggerate local disparities and weaken their identity across all levels.

Comparing the clusters, the Institutional–Performance Cluster showed a mean total mediation effect of 0.260, approximately 2.4 times the 0.108 observed for the Emotional–Traditional Cluster, with Party Identity at 0.136 falling in between. The mean political participation pathway effect for the institutional cluster was 0.107, far exceeding the 0.043 of the emotional cluster. The author concludes that institutional and governance-related identities are more readily fostered through digital channels, while emotional and party identities depend more heavily on real-world experiences such as rights protection and public services, so purely online information campaigns yield limited results for those deeper attachments.

The study frames its findings as qualifying the universal claims of Western polarization theory: at the aggregate level, perceptions of digital technology predicted all eight identity dimensions positively, and no general erosion or fragmentation of political identity appeared in this sample. At the same time, the results do not support a simple neutrality thesis, because effects flowed through five distinct mediators, including a consistently negative one. The author acknowledges key limitations: the sample consists solely of college students, limiting generalization to the broader young adult population, and the cross-sectional design cannot establish dynamic, time-ordered causal relationships. Future work, the paper suggests, should extend to young adults in the general population and new economic sectors, add longitudinal tracking, and test moderators such as digital usage duration and media trust.

Subject of Research: How perceptions of digital and intelligent technologies mediate multidimensional political identity among Chinese college students

Article Title: The heterogeneous effects of perceptions of digital and intelligent technologies on multidimensional political identity among college students: a study of multiple parallel mediations from a stratified cluster perspective

Article References: Hou, Q. (2026). The heterogeneous effects of perceptions of digital and intelligent technologies on multidimensional political identity among college students: a study of multiple parallel mediations from a stratified cluster perspective. Frontiers in Psychology, 17, Article 1794810. https://doi.org/10.3389/fpsyg.2026.1794810

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1794810

Keywords: political identity, digital technology, college students, mediation analysis, political socialization, algorithmic polarization, political participation, China, survey research, media psychology, bootstrap mediation, youth