TL;DR
This study tracks how Chinese university students perceived the risk of COVID-19 over a 20-month period, combining Latent Profile Analysis (LPA) and Network Analysis. The researchers found that risk perception significantly declined over time, driven by a "psychological immunization" effect and high trust in government interventions, while identifying two distinct psychological subgroups of students.
What Problem Does This Paper Address?
Understanding how people perceive risk is vital for public health; if people don't feel "at risk," they are less likely to follow safety protocols like vaccination or masking. While many studies have looked at COVID-19 risk perception, most provide only a "snapshot" in time or treat the population as a single, homogenous block.
The authors address two major gaps:
- Subgroup Identification: Do all students perceive risk the same way, or are there hidden "profiles" (e.g., a high-anxiety group vs. a stoic group)?
- Structural Dynamics: How do different facets of risk—like personal worry versus concern for the country—interact with each other, and does that "network" of thoughts change as a pandemic drags on?
Methods
The study employed a sophisticated dual-statistical approach:
- Latent Profile Analysis (LPA): Instead of just averaging scores, LPA identifies hidden subgroups (latent classes) within the data based on similar response patterns across eight risk-related questions.
- Network Analysis (EBICglasso): The researchers modeled risk perception as a system of interacting "nodes" (survey items). They used the Extended Bayesian Information Criterion Graphical Least Absolute Shrinkage and Selection Operator (EBICglasso) to estimate the strength of connections (edges) between these nodes, identifying which specific concerns were central to the students' overall mindset.
- Longitudinal Tracking: Data was collected at three critical junctures: January 2020 (the initial outbreak), January 2021, and September 2021.
Datasets
The study utilized a primary dataset collected via the Wenjuanxing online platform:
- Cross-sectional sample: 1,837 Chinese university students from four provinces (Jiangxi, Heilongjiang, Shannxi, and Liaoning) at T1.
- Longitudinal sample: A subset of 334 students who completed the survey at all three time points (T1, T2, and T3).
- Instrument: An eight-item risk perception questionnaire adapted from previous literature, measuring both individual risk (e.g., "I have chances of contracting COVID-19") and public risk (e.g., "The outbreak affects the whole country").
Key Insights and Ideas
- The "Two-Class" Reality: The population wasn't a spectrum; it fell into two distinct buckets. Roughly 54.7% of students belonged to a "low risk perception" class, while 45.3% were in a "high risk perception" class.
- National Concern as a Hub: In the network analysis, the node representing the outbreak's effect on the whole country (rp5) had the highest "strength centrality." This suggests that for these students, their personal sense of risk was deeply tied to their perception of the national situation—a hallmark of a collectivist culture.
- Psychological Immunization: Risk perception didn't just fluctuate; it decreased continually from 2020 to late 2021. The authors attribute this to "psychological immunization"—repeated exposure to the threat and the rollout of vaccines (over 2.2 billion doses in China by late 2021) reduced the "dread" factor.
- Stability of the Network: Interestingly, while the intensity of fear dropped, the structure of the risk network remained stable. The way different concerns (like personal worry vs. perceived probability) related to one another didn't fundamentally change, even as the overall level of concern fell.
Key Results
- Quantitative Decline: Risk perception scores dropped significantly across all three time points ($p < 0.05$).
- Strongest Links: The strongest connection in the risk network was between "average people have chances of contracting" and "average people worry about catching" ($r = 0.491$).
- Gender Nuance: While some studies suggest women perceive higher risk, this study found no significant gender differences in the overall latent classes at the start of the pandemic, though some specific items showed minor variations later on.
- Centrality: Node rp5 (national impact) was the most influential "hub" in the cross-sectional network, meaning it was the strongest predictor of other risk perceptions.
Limitations and Future Work
The authors acknowledge several constraints:
- Convenience Sampling: The participants were from specific regions and may not represent the entirety of the Chinese university student population.
- Self-Reporting Bias: Survey data is subject to subjectivity and may lack the objectivity of behavioral data.
- Missing Links to Behavior: While the study maps perception, it doesn't directly measure how these changing perceptions translated into actual behaviors (like mask-wearing or social distancing) in the later stages of the pandemic.
Future research could benefit from using Receiver Operating Characteristic (ROC) curves to establish objective "thresholds" for high-risk profiles and exploring how these psychological networks interact with social media consumption.
Summary updated