Introduction

Political crises often reshape citizens’ political attitudes and behavior by heightening perceptions of threat and uncertainty [1,2]. A large body of research shows that during crises, citizens tend to rally behind political leaders and incumbent governments, a phenomenon commonly referred to as the rally-around-the-flag effect [3,4]. This effect has been most consistently observed in response to security-related crises and public health emergencies, and is typically attributed to heightened threat perceptions, emotional responses such as fear and anxiety, and demands for decisive leadership [5,6].

However, recent scholarship suggests that rally-around-the-flag effects are often time-bound rather than permanent [7,8]. As crises persist, initial surges in trust and incumbent support may attenuate at the aggregate level, giving way to performance-based evaluations, economic concerns, and renewed political contestation [9]. In the context of the COVID-19 pandemic, research shows that as emergency conditions became normalized, aggregate rally-around-the-flag effects weakened [8]. Importantly, the attenuation of aggregate rally-around-the-flag effects does not imply that crisis-related perceptions become politically irrelevant. Research on COVID-19 demonstrates that perceived risk, uncertainty, and crisis-related anxiety often persist beyond the acute phase and continue to shape political attitudes and evaluations, even as objective health risks decline [10,11]. Recent studies further show that the political consequences of the pandemic have extended into the post-emergency period, influencing institutional trust, democratic attitudes, and policy preferences across different national contexts [12,13]. These findings raise new questions about whether and how underlying mechanisms such as risk perception and crisis-related government performance evaluations continue to influence political behavior during the later stages of prolonged crises, when uncertainty remains, even after the most acute phase of the crisis has passed.

South Korea provides an analytically informative case for examining how prolonged crises shape political behavior beyond the initial rally-around-the-flag phase because it experienced two nationwide elections at distinct stages of the same crisis. Research on crisis politics demonstrates that emotional responses to threat and uncertainty can influence electoral behavior across diverse democratic contexts, even when crises are triggered by different events and unfold under varying institutional arrangements [14,15]. In South Korea, the early phase of the COVID-19 pandemic coincided with the April 2020 National Assembly election, during which electoral outcomes were consistent with a pronounced rally-around-the-flag effect associated with the government’s pandemic response [16]. Evidence from the April 2020 legislative elections further indicates that evaluations of the government’s COVID-19 countermeasures became increasingly consequential for vote choice as the pandemic progressed.

By contrast, the later stage of the pandemic culminated in the March 2022 presidential election, held after crisis management had become routinized and acute emergency pressures had largely subsided. Analyses of this election emphasize the reassertion of conventional electoral mechanisms, including strong partisan polarization, habitual partisanship, and retrospective evaluations of the incumbent administration, rather than crisis-driven rally-around-the-flag effects [17,18]. This temporal sequencing—an early election marked by rally-around-the-flag effects followed by a later election dominated by conventional electoral mechanisms—provides a suitable analytical context for examining how risk perceptions and evaluations of government performance (overall, pandemic response, and economic) relate to voting intentions as rally-around-the-flag effects attenuate over the course of a prolonged crisis.

Building on this context, we examine voter behavior in South Korea’s March 2022 presidential election to assess how risk perceptions, evaluations of the government’s handling of the COVID-19 pandemic, and economic assessments shaped voting intentions during the later stages of the prolonged crisis. Situating the analysis within the literature on crisis politics, this study moves beyond the identification of rally-around-the-flag effects to investigate how crisis-related perceptions and retrospective evaluations continue to operate once acute emergency pressures have subsided and electoral competition has resumed under routinized crisis management.

Literature review

Research questions and conceptual framework

To shed light on the association between risk perception and voting intention during the final phase of the COVID-19 pandemic, the following research questions (RQs) are addressed using survey data collected during the 2022 presidential election in South Korea:

RQ1. How did individuals’ risk perceptions relate to their intention to vote for the incumbent party’s candidate in an election held at the end of the COVID-19 pandemic?

RQ 2. What was the relationship between voters’ evaluation of the government’s performance and their intention to vote for the incumbent party’s candidate in an election held at the end of the COVID-19 pandemic?

Fig 1 illustrates the study’s conceptual framework, showing how political responses evolve across different phases of a prolonged crisis, from rally mobilization during the acute phase to pre-crisis political dynamics as crisis management becomes routinized. The framework also clarifies that both RQs are examined within a shared empirical context using data from a single cross-sectional survey.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Conceptual framework of crisis phases and political responses. Note: RQ1 = research question 1; RQ2 = research question 2.

https://doi.org/10.1371/journal.pone.0345621.g001

Materials and methods

Results

As shown in Table 1, slightly more than half of the respondents were men (52.3%), while white-collar workers (31.3%) and full-time homemakers (16.5%) collectively accounted for slightly less than half of the sample. Nearly half of the sample identified their household economic status as middle class (49.5%), and a similar proportion lived in a metropolitan area (49.3%). Most respondents reported a college education or higher (65.3%). Slightly less than half of the sample approved of the incumbent government’s overall performance (48.1%), and approximately half of the sample approved of its handling of the COVID-19 pandemic. With respect to COVID-19 risk perception, more respondents were worried (21.4% very worried and 37.8% somewhat worried) than not worried (26.0% not really worried and 14.8% not worried at all), and expected to be infected (22.1% very likely and 53.4% somewhat likely) than not (16.8% not really likely and 7.7% not at all likely). Regarding economic conditions, a majority judged the national economy to have worsened over the preceding year (52.4%), while most indicated that their household economic situation had remained similar (57.5%). Expectations for both national (41.4%) and household (55.2%) economic conditions were most commonly that conditions would remain the same.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics and odds ratios between explanatory variables and voting intention.

https://doi.org/10.1371/journal.pone.0345621.t001

Table 1 summarizes the results of the weighted chi-squared tests and the weighted logistic regression analyses addressing RQ1 and RQ2. The chi-squared tests examine the marginal associations between voting intention and each explanatory factor, indicating that voting intention differed significantly across most variables, with the exception of self-reported household economic status, educational attainment, and cognitive risk perception.

The logistic regression results are reported as adjusted odds ratios with 95% confidence intervals in parentheses. The reference category for the outcome variable (voting intention) is the main opposition candidate (i.e., Suk-yeol Yoon).

First, cognitive risk perception was positively correlated with voting intention for the incumbent party’s candidate. Although it was not clear ex ante whether the rally-around-the-flag effect would persist in the end phase of the COVID-19 pandemic, this result is descriptively consistent with patterns discussed in the literature.

For government approval, both overall approval and countermeasures to the COVID-19 pandemic were positively correlated with intention to vote for the incumbent party’s candidate. This finding was consistent with the aforementioned retrospective voting framework. Similarly, most variables related to the economic situation were positively correlated with intention to vote for the incumbent party’s candidate. Specifically, those who judged the national economic situation as similar or improved were more likely to vote for the incumbent party’s candidate than those who considered it to have worsened. Similarly, those who expected the national economy to remain the same or improve had a higher intention to vote for the incumbent party’s candidate than those who expected it to worsen. Moreover, respondents who expected their households’ economic situation to remain the same were more likely to vote for the incumbent party’s candidate than those who expected it to worsen.

Some socioeconomic characteristics were also significantly correlated with voting intention. For instance, women were more likely to vote for the incumbent party’s candidate, while students were less likely to vote for the incumbent candidate than those who were “unemployed/retired/other.” Those living in Honam (southwest of the country) had higher intentions than those living in metropolitan areas.

Discussion

This study uses survey data from the 2022 South Korean presidential election to examine how voting intentions during the later phase of the COVID-19 pandemic were associated with different dimensions of risk perception and broader evaluations of government and economic performance.

First, in the end phase of the COVID-19 crisis, cognitive risk perception showed a small but significant association with voting intention in the South Korean presidential election, whereas affective risk perception did not (RQ1). This contrasts with existing findings from earlier phases of the pandemic suggesting that affective risk plays a critical role in increasing the trust and approval of incumbent leaders, often triggering a rally-around-the-flag effect in response to heightened uncertainty [16,53]. The attenuated relationship between risk perception and voting intention observed in our study may reflect a broader decline in perceived risk as the pandemic persisted, and individuals adapted to the ongoing threat [54]. This is in line with prior research showing that prolonged exposure to crisis conditions can lead to desensitization or “risk fatigue,” especially when individuals adapt to the threat and view it as manageable [20,36]. This pattern is also reflected in the 2023 Gallup Korea public opinion poll; while the overall approval of the Moon Jae-in administration’s pandemic response remained relatively high, support for the ruling party’s presidential candidate declined in the final months before the election [55]. In other words, voters’ evaluations of COVID-19 risk and crisis management changed by the end of the pandemic.

One possible explanation, consistent with the literature on crisis normalization, is that everyday life under COVID-19 was increasingly perceived as routine during this period. South Korea’s early and sustained success in containing outbreaks, transparent public health messages, and the institutionalization of protective behaviors (e.g., QR code check-ins, universal mask-wearing) may be interpreted as coinciding with a widespread perception that the crisis was under control [56]. In the later phase of the pandemic, affective risk perception was less strongly associated with voters’ evaluations of the incumbent and their responses to the pandemic. However, among the smaller group of voters reporting relatively high levels of attentiveness to infection trends and policy performance, cognitive risk perception was more salient in its association with voting preferences, particularly in support for the incumbent party, within a broader context of positive evaluations of South Korea’s pandemic management. The Social Amplification of Risk Framework [35,36] suggests that risk perceptions are shaped not only by the objective characteristics of a threat but also by how institutions communicate and manage that threat over time. Even as COVID-19 became largely routine, continued exposure to official messaging, media framing, or personal experiences may help explain why perceived risk remained salient in some individuals’ electoral decision-making [31,57,58].

Second, as the COVID-19 pandemic waned, voting intentions were associated not only with approval of the government’s pandemic response but also with conventional performance evaluations, including overall government approval and economic evaluations and expectations (RQ2). These findings indicate that, during the later stage of a prolonged crisis, voting intentions were not structured primarily around pandemic-related approval but were instead associated with a broader set of performance-related considerations. This pattern can be situated within the broader literature on crisis politics. Prior research has consistently shown that rally-around-the-flag effects are inherently time-bound and tend to attenuate as crises become prolonged [7,8,40,41]. In the context of the COVID-19 pandemic, early electoral outcomes in multiple democratic settings—including South Korea’s April 2020 National Assembly election—were widely interpreted as being consistent with rally-around-the-flag dynamics associated with approval of governments’ pandemic responses [16]. These early patterns were linked to heightened uncertainty, perceived threat, and the political salience of crisis management during the acute phase of the pandemic.

In the South Korean context, pandemic governance was characterized by centralized coordination, extensive digital contact tracing, and sustained public communication, which contributed to the institutionalization of COVID-19 management as a routinized policy domain rather than an acute emergency [59–61]. Against this background, the associations observed in the present study are consistent with expectations derived from the crisis politics literature. Rather than suggesting that pandemic-related considerations disappeared from electoral judgment, the findings indicate that evaluations of COVID-19 countermeasures operated alongside other indicators of government performance in the later stage examined here. While the present study does not directly assess changes in rally-around-the-flag dynamics, this pattern is consistent with prior studies suggesting that the political salience of crisis-specific considerations diminishes over time as performance-based evaluations regain prominence [7,8,40,41].

These economic and general performance evaluations did not occur in isolation from the crisis [48,49,62]. Rather, they were observed alongside the broader consequences of prolonged public health restrictions, including lockdown-related economic disruptions, job insecurity, and fiscal relief efforts [63,64].

From this perspective, the present findings align with existing research demonstrating that evaluations of COVID-19 countermeasures retained political relevance beyond the acute phase of the pandemic [10–13,40,50,65]. Studies conducted across national contexts have shown that perceptions of pandemic response, perceived risk, and crisis-related anxiety continued to be associated with political attitudes and electoral preferences even as objective health risks declined [10–13,48,49]. In line with this literature, approval of the government’s pandemic response remained significantly associated with voting intention in the present analysis, indicating that pandemic-related evaluations continued to form part of voters’ political considerations.

Taken together, the findings indicate that during the later stage of the COVID-19 pandemic, voting intentions were associated with a configuration of performance evaluations that more closely resembles conventional retrospective patterns than crisis-centered ones. By focusing on the later phase of the pandemic, the present study extends previous work that has concentrated primarily on early crisis dynamics [16,40,41] and contributes to a more temporally differentiated understanding of how crisis-related and conventional evaluations coexist in shaping electoral behavior.

Finally, several limitations of this study should be noted. First, reliance on a pre-election survey without measures of voting intention strength limits assessment of how stated preferences translated into actual electoral behavior. Second, because the study is cross-sectional, the findings do not allow any causal interpretation or assessment of changes over time, underscoring the need for future longitudinal and qualitative research. Third, the distinction between cognitive and affective risk perception is subject to measurement limitations, as the survey design did not allow for sufficient ex ante consideration of design features such as randomization of question order. Finally, the findings should be interpreted in light of South Korea’s institutional context, which offers a particularly informative case for understanding how electoral behavior unfolds during a prolonged public health crisis.

This study addresses the need to better understand electoral behavior during the later phase of a prolonged crisis, a period that has received far less scholarly attention than the acute onset of the COVID-19 pandemic. By examining how risk perceptions and performance evaluations relate to voting intentions at this stage, the study advances a more temporally differentiated perspective on crisis politics and electoral accountability. Academically, the study contributes by clarifying the distinct role of cognitive risk perception and by showing that evaluations of pandemic policies coexist with conventional assessments of government and economic performance in electoral judgment during a prolonged crisis. Practically, the findings imply that during extended pandemic conditions, governments cannot rely solely on early crisis management or public health achievements but must also demonstrate effective responses to ongoing economic and political challenges and communicate these efforts clearly to voters as active evaluators.