The Lasting Effects of the COVID-19 Pandemic on Support for Anti-Democratic Political Systems: A Six-Month Longitudinal Study.

Objectives: We aimed to analyze the enduring effects exerted by COVID-19 exposure and subjective economic threats on support for anti-democratic political systems. Method: We performed a two-wave longitudinal study on a quota panel of the Italian adult population (N  =  1,073), surveyed first in April 2020, during the first peak of the pandemic (T 1) and the first lockdown, and second in October 2020, at the onset of the second peak of the pandemic and of the second lockdown (T 2).
Results: A mediation model showed that COVID-19 exposure and subjective economic threats, measured at T 1, fostered a negative evaluation of how the government managed the pandemic, which, in turn, had a positive association with support for anti-democratic political systems, both measured at T 2.
Conclusion: The existential threats related to the COVID-19 pandemic have enduring political consequences, undermining people's support for democracy, even six months after their onset.

In fascinating archival research performed using territorial data of German towns in the 1910s–1930s, Blickle (2020) recently showed a strong association between the number of influenza deaths in 1918 and the increased number of votes for the Nazi Party in the 1932 and 1933 elections. The findings from two recent studies performed with individual data in the context of the COVID‐19 pandemic converge with Blickle's results, showing that large‐scale pandemics can jeopardize support for democratic institutions and systems (see Amat et al., 2020; Roccato et al., 2020).

The results from the three studies described above are consistent with the affective intelligence theory (e.g., Marcus, Neuman, and MacKuen, 2000) and with the social‐psychological literature on the relations between threats and authoritarianism (e.g., Onraet et al., 2013). The affective intelligence theory conceives emotions as adequate responses to the significance that particular circumstances hold for an individual. In this light, emotions regulate cognitive resource allocation and function as signals directing attention to potential environmental opportunities or threats (Damasio, 1994; Johnston, 1999). Accordingly, far from being irrational (Demertzis, 2013), emotions are regarded as an integral component of an effective decision‐making process (Bonansiga, 2020). In light of the affective intelligence theory, threatening situations—such as pandemics—activate people's surveillance systems. The emotional expression of such activation is anxiety, which signals the presence of problems in relation with the external world and the need for careful and systematic information processing to achieve a fuller understanding and analysis of the threat's nature. Consistent with this, political psychology research has shown that anxiety is often associated with the development of new political attitudes and even with changes of voting behaviors (Marcus and MacKuen, 1993).

The literature on the relations between threats and authoritarianism shows that people can react to threats by resorting to secondary sources of control, with the function of compensating the loss of direct control over their life and environment (e.g., Kay et al., 2008). Before COVID‐19, the literature mainly focused on the existential threats stemming from natural disasters (e.g., Russo et al., 2020), climate disasters (e.g., Zapata, 2018), and economic crises (e.g., Chen, 2010), showing the role of compensatory sources of control played by religion (e.g., Sibley and Bulbulia, 2012), the government (e.g., Oneal, Lian, and Joyner, 1996), and, most importantly for this study, anti‐democratic authorities (Mirisola et al., 2014). In line with this, more recent studies indicated that the COVID‐19 pandemic has led to an increase in religiousness and church attendance (e.g., Molteni et al., 2021), trust in the government (e.g., Bol et al., 2021), and anti‐democratic attitudes and preferences (e.g., Amat et al., 2020; Roccato et al., 2020). Amat et al. (2020), using a mix of experimental and survey data, showed that in Spain the COVID‐19 pandemic has fostered a widespread demand for strong leadership and even a willingness to tackle the crisis by giving up some individual freedom. In a longitudinal study performed in Italy, Roccato et al. (2020) showed that COVID‐19 exposure and perceived economic threats were positively associated with support for anti‐democratic political systems, independent of participants’ prepandemic predisposition toward them. But how long does this effect persist? Answering this interesting question will help to define the sociopolitical impact of the pandemic, beyond its health consequences, as a precondition to successfully face it in the near future.

The COVID‐19 Pandemic in Italy

Italy was the first European country to be drastically hit by the COVID‐19 pandemic. During the first stage of the virus assault, the daily death toll peaked at 919 (March 27, 2020). The Italian government dealt with the emergency through a radical lockdown strategy. The first decree law by the Italian prime minister (D.P.C.M., March 8, 2020) established that citizens could only leave their homes for documented work activities, health reasons, and other emergencies. Schools, universities, and all nonessential shops and industries were closed, all public events cancelled, and travel drastically restricted. This first national lockdown period (March 9, 2020−May 18, 2020) helped reduce the death toll to below 100 per day, starting on 25 May.

When we collected the data for our second survey, six months after the first wave of the pandemic and the first lockdown, the situation remained far from normal. Indeed, a second pandemic wave was at its onset, and the Italian government tackled it via new lockdown measures, even if less restrictive than those introduced in March–April 2020. When we wrote this article (January 2020), Italy had passed the thresholds of two million cases of COVID‐19 and 75,000 deaths due to the virus.

The Present Study

At least two main questions are still to be answered in this field of study. First, we do not know the long‐term effects of the COVID‐19 pandemic. In Blickle's (2020) analysis there is a gap of nearly 15 years between the cause (the number of deaths due to influenza in 1918) and its studied effect (increased votes for the Nazi party in 1932 and 1933). However, the world has undoubtedly changed from the 1920s and 1930s. As travel and communication are incommensurably faster today than they were a century ago, it is plausible that the time duration before the onset of social‐psychological effects of a pandemic is reduced. Additionally, results from studies using territorial data, such as Blickle's (2020), can be distorted by the ecological fallacy: associations that hold at the contextual level do not necessarily reflect analogous associations among individuals belonging to those contexts (Robinson, 1950). Thus, we believe that an acritical generalization of Blickle's (2020) results would be rash, and that contemporary analyses performed on individual data are needed to uncover the duration of the pandemic effects. The second limitation of the extant literature is that we do not know the processes through which existential threats lead to support for anti‐democratic political systems. As we will detail next, we reasoned that the evaluation of how the government tackled the pandemic could be a promising candidate mediator of the link between existential threats and support for anti‐democratic political systems.

In this study, we tested a six‐month longitudinal mediation model aimed at predicting Italians’ support for anti‐democratic political systems as a function of their COVID‐19 exposure and subjective economic threats, with the mediation of the evaluation of how the government managed the pandemic from the public health and economic points of view.

Based on the literature above, it is far from surprising that, just like the rest of the world, the COVID‐19 pandemic has fostered a severe sense of unpredictability among Italians, due to their need to cope with a completely novel and unknown threat (Molteni et al., 2021). In addressing the effects of the existential threats posed by COVID‐19, we investigated the mediating role of the evaluation of the government's tackling of the pandemic. We expected a stronger dissatisfaction among those who had higher COVID‐19 exposure and showed higher levels of subjective economic threat due to their more extensive experience of the many inefficiencies of the management of the health and economic procedures as systematically expressed in opinion polls (e.g., 〈https://www.investireoggi.it/news/sondaggi‐politici‐e‐covid‐al‐21‐ottobre‐gli‐italiani‐preferiscono‐conte‐e‐le‐sue‐restrizioni/〉) and the mass media (e.g., 〈https://www.ilmessaggero.it/editoriali/primopiano/covid_fuori_controllo_il_prezzo_alto_da_pagare_per_gli_eccessi_dell_estate‐5540177.html〉). Consistent with this, public opinion polls showed that, after a short‐term honeymoon phase for the government, after summer 2020, Italians began to develop more negative evaluations of how the government managed the pandemic. For instance, an IPSOS poll showed that the measures the government took to limit the spread of the pandemic were evaluated positively by 62 percent of Italians in March, but only by 45 percent in November 2020. The measures taken to support families were evaluated positively by 55 percent of Italians in March, but only by 24 percent in October (see 〈https://www.corriere.it/cronache/20_novembre_18/vaccino‐covid‐un‐italiano‐sei‐rifiutera‐farsi‐vaccinare‐4dbfdb82‐2924‐11eb‐92be‐ccd547aa4d2b.shtml〉). Accordingly, in our data set, the percentage of participants who considered the government's measures adequate for managing the pandemic fell significantly between April (42.04 percent) and October 2020 (35.48 percent), χ2(1) = 8.43, p = 0.004. Thus, if the first onset of COVID‐19 was perceived as a totally unpredictable event for which the government could not be blamed, the second onset induced greater public criticism, with many contending that more could be have done to prevent it.

Overall, we hypothesized that COVID‐19 exposure (H1a) and subjective economic threats (H1b) foster a negative evaluation of government performance in response to the pandemic. We also expected a positive association between a negative evaluation of government performance in response to the pandemic and support for anti‐democratic political systems (H2), in that the preference for an anti‐democratic political system can be used as a source of compensatory control over a threatening and uncontrollable world, particularly after the perceived failure of democratic institutions in managing the pandemic. Although the data do not allow for straightforward causal inference, we nonetheless argue that the general context in which the surveys were held and the analytical strategy applied lead to findings of very high relevance for understanding the potential impacts of the COVID‐19 crisis on the support for anti‐democratic attitudes.

Method

Participants and Procedure

We surveyed via web a quota panel of the Italian adult population, stratified by gender, age, geographical area of residence, and size of area of residence (more information is available on the CoCo project website, 〈https://www.dippsicologia.unito.it/do/progetti.pl/Show?_id=9fxo〉). We collected the data between April 17 and 26, 2020 (T 1¸ N = 1,199) and between October 9 and 28, 2020 (T 2, N = 1,151). We had complete data for both waves from 1,073 respondents (51.3 percent women; M age = 49.23; SD = 14.32). The participants are members of a panel systematically interviewed by the Bilendi research institute (see 〈https://www.bilendi.co.uk/〉). Thus, this low dropout rate is not surprising. The gender distribution of the sample overlapped that of the Italian adult population (51.3 percent women). However, our sample was slightly older than the Italian population (M age  =  45.7).

A logistic regression showed that the dropout from T 1 to T 2 (dropout = 0, retention = 1) was not associated to participants’ gender, education, exposure to COVID‐19, or perceived economic threat. A small positive effect of age was detected (Wald = 9.66, p = 0.002): as age increased, the respondents were more inclined to participate also in the second wave. However, a low Nagelkerke R 2 (0.02) confirmed that the differences between those who participated in the two waves and those who participated only in the first one were not substantial.

Measures

Independent Variables

We measured participants’ exposure to COVID‐19, both at T 1 and T 2, using the following item: "Did you or some person close to you (such as a relative or close friend) contract COVID‐19?" Response alternatives were No (=1), Not me, but some people close to me did (=2), Yes, I did, but none of the people close to me did (=3), and Yes, both me and some of the people close to me (=4). We measured participants’ perceived economic threat, both at T 1 and T 2, using the following European Social Survey item: "Which of the following descriptions comes closest to how you feel about your household's income nowadays?" Response alternatives were Living comfortably on present income (=1), Coping on present income (=2), Finding it difficult on present income (=3), and Finding it very difficult on present income (=4) (cf. Roccato et al., 2020). To test our hypotheses, we included both T 1 measures as a baseline and their change between T 1 and T 2. For COVID‐19 exposure, we have combined those who had directly or indirectly (when some people close to them contracted the virus) contracted COVID‐19. Based on this, we computed an indicator of new exposure, contrasting those who had virus exposure after April 2020 (=1) versus those who had virus exposure before April 2020 or had no virus exposure at all (=0). Concerning the economic threat, we computed a change score by subtracting the perceived economic threat at T 1 from the perceived economic threat at T 2 (Δ economic threat).

Mediator Variable

At T 2, we asked participants about their evaluation of how the government managed the pandemic, using two items developed ad hoc for this research: "How satisfied are you regarding the decisions made by the government to tackle the spread of the COVID‐19 pandemic?" and "How satisfied are you regarding the decisions the government has made to sustain and revamp the economy?" Response alternatives were Not at all (=1), A little (=2), Somewhat (=3), and Extremely (=4). The two items were highly correlated, r = 0.726, p < 0.001.

Dependent Variable

We measured participants’ support for anti‐democratic political systems at T 2 using the following items: "I am going to describe various types of political systems and ask what you think about each as a way of governing this country in a period such as the present. For each one, would you say it is a very bad ( = 1), fairly bad ( = 2), fairly good ( = 3), or very good ( = 4) way of governing this country? Having a strong leader who does not have to bother with parliament and elections’ and ‘Having the army rule’ (the same items are used in the World Values Survey). The correlation between the two items was r = 0.503, p < 0.001.

Control Variables

In our model, we controlled for gender (1  =  woman), age, years of education, and occupational status (0 = employed, 1 = unemployed).

Data Analysis

Using MPLUS 8 (Muthén and Muthén, 1998–2017), we tested our hypotheses via a structural equations model. We modeled the mediator and the dependent variable as latent variables. In our model, COVID‐19 exposure and the perceived economic threat at T 1, as well as their change between T 1 and T 2, predict satisfaction with government decisions (measured at T 2), which, in turn, is associated with support for anti‐democratic systems (T 2). The model is depicted in Figure 1.

FIGURE 1

The Model Tested

We evaluated the fit of our model using the χ2, the comparative fit index (CFI), Tucker–Lewis index (TLI), and the root mean square error of approximation (RMSEA). Based on Hu and Bentler (1998), we considered the models satisfactory with a TLI and a CFI > 0.90 and with a RMSEA < 0.05 (Xia and Yang, 2018).

Results

Table 1 reports the correlations among the variables we used. The model fit was excellent: χ2 (21) = 43.491, p = 0.003, CFI = 0.981, TLI = 0.966, RMSEA = 0.032, with 90 percent CI: 0.018–0.045. As reported in Table 2, respectively, consistent with H1a and H1b, both COVID‐19 exposure at T1 and the perceived economic insecurity at T 1 reduced the satisfaction with government decisions at T 2. However, only the change in economic insecurity between T 1 and T 2, but not new contacts with COVID‐19, additionally reduced satisfaction with the government. Consistent with H2, the latter had a negative association with support for anti‐democratic systems.1

TABLE 1

Correlations Among the Study Variables

	2	3	4	5	6	7	8	9	10
1. COVID‐19 exposure T 1	−0.180***	−0.017	−0.002	−0.093**	0.046	−0.102***	−0.049	−0.141***	0.086**
2. New COVID‐19 exposure	–	−0.056	−0.009	0.078*	0.071*	−0.094**	0.029	−0.071*	0.061*
3. Economic threat T 1		−	−0.447***	−0.136***	0.049	0.043	0.098***	−0.005	−0.130***
4. Δ economic threat			–	−0.074*	0.007	0.057	0.039	0.031	−0.013
5. Satisfaction with government's decisions				−	−0.099**	−0.068*	−0.054	0.009	0.084**
6. Authoritarian political preferences					−	−0.075*	−0.002	−0.113***	−0.073*
7. Unemployment						–	0.215***	0.318***	−0.299***
8. Gender							−	0.028	−0.104***
9. Age								−	−0.248***
10. Education									–

TABLE 2

Results from the Path Analysis: Standardized Coefficients

	Mediator: Satisfaction with Government's Decisions		Dependent Variable: Authoritarian Political Preferences
	β	p	β	p
Gender	−0.021	0.527	0.002	0.964
Age	0.030	0.386	−0.141	0.007
Education	0.062	0.073	−0.140	0.001
Unemployment	−0.048	0.185	−0.103	0.016
COVID‐19 exposure T 1	−0.103	0.002	−	–
New COVID‐19 exposure	0.042	0.199	–	–
Economic threat T 1	−0.216	<0.001	–	–
Δ economic threat	−0.165	<0.001	–	–
Satisfaction with government's decisions	–	–	−0.113	0.006
R 2	0.071		0.056

Discussion

The COVID‐19 pandemic, one of the largest exogenous shocks of the century (Wiwad et al., 2021), poses an inherently global existential threat with no delimitated enemy (Hartman et al., 2021). Previous research shows that dramatic events such as pandemics can undermine the social order, sparking fear, panic, stigma, and large‐scale calls to action (Strong, 1990). However, they can also inspire acts of unity, compassion, and solidarity (e.g., Ntontis and Rocha, 2020). These contrasting results are not difficult to reconcile, as we can consider pandemics as exogenous shocks with potentially destabilizing consequences for the status quo (Amat et al., 2020).

The COVID‐19 pandemic has stimulated a stream of studies analyzing its individual and social consequences, mainly, if not exclusively, focusing on its short‐term effects. To our knowledge, generalized results in the medium term are still scant (for an exception, see Blickle, 2020). Additionally, to date, no study has investigated the processes through which existential threats lead to anti‐democratic orientations. This work contributes to this literature, attempting to address this gap of knowledge by focusing on medium‐term sociopolitical impacts, exploring how the existential threat triggered by COVID‐19 induced the activation of compensatory sources of control in terms of support for anti‐democratic systems. We showed that the participants who were directly exposed to COVID‐19 and experienced a subjective economic threat were also those who tended to have a negative evaluation of the government's management of the crisis and, in turn, tended to look at anti‐democratic systems for hope, even six months after the onset of the pandemic. Thus, the COVID‐19 pandemic shock proved to be sufficiently strong to foster enduring effects (Ashworth, Bueno de Mesquita, and Friedenberg, 2018).

Our results are consistent with the literature showing that the COVID‐19 pandemic strengthened anti‐democratic attitudes and preferences (e.g., Amat et al., 2020; Roccato et al., 2020), and it adds to it, showing that this authoritarian shift is not a contingent short‐term effect. On the contrary, the function of an anti‐democratic orientation to compensate for the perceived loss of control due to the existential threat persisted at least for several months. Additionally, our findings revealed the processes through which the existential threat (both in health and economic indicators) led to support for anti‐democratic political systems. The significance of our mediation model is in line with previous studies on the rally effect, indicating that the ability to assess the performance of institutions in managing a crisis is a key factor in shifting public opinion after exogenous shocks (e.g., Colaresi, 2007). Our results are consistent with this approach, although the compensatory control mechanism we considered was not support for the government, but, conversely, support for anti‐democracy.

The present study has some limitations that need acknowledgment. Although the longitudinal nature of our findings is a strength of the current research, we measured our mediator and dependent variable at the same time. Future research should establish the generality of the mediated effects reported in the current investigation by conducting longitudinal studies with multiple waves. Additionally, the analysis of the extent to which similar effects are fostered by actual health and economic data could represent another development of this study. Finally, our data focus on the COVID‐19 emergency only. The social‐psychological literature on authoritarianism shows strong links between societal threat and anti‐democratic attitudes (e.g., Onraet et al., 2013). However, such effects are typically studied in a short‐term perspective. We are confident about the generalizability of our results to other kinds of societal threat. However, an explicit test of the generalizability of our model to other emergencies could be interesting.

Despite the above‐mentioned limitations, the present research offers novel results showing that studying the COVID‐19‐related medium‐term impact is of utmost importance. The first survey was collected during the first wave of the COVID‐19 pandemic, in the middle of the first lockdown, whereas the second survey was conducted at the beginning of the second pandemic wave. Our longitudinal approach is potentially suitable to study socio‐psychological changes fostered by COVID‐19 as the result of individual trajectories. Based on the evidence provided in this study, it would be interesting to undertake periodic monitoring of changes in public opinion in relation to the progress of the pandemic and to its management, until its spread and its health and economic consequences have been reduced on a large scale. Beyond the study of this specific pandemic, this approach could be useful in developing new models for the analysis of public opinion. In fact, almost exactly 100 years after the Spanish flu—the previous large‐scale pandemic that spread in a very different and less‐connected world—the COVID‐19 pandemic represents a (hopefully) once‐in‐a‐lifetime opportunity to investigate how society functions and acts during global crises. Finally, evidence relating to the mediation effect suggests that the anti‐democratic consequences of the pandemic can be context‐specific, depending on the role played by institutional performance of government. In particular, it remains to be seen whether this empirical model is cross‐culturally generalizable to a large number of Western democracies, also analyzing potential context‐specific predictors of our mediation. A test of the generalizability of our model in countries where the pandemic was managed more effectively than in Italy (e.g., in Australia: see Stanaway et al., 2021) or, as in the “Swedish experiment” (see Esaiasson et al., 2020), by resorting to much less draconian measures would be particularly interesting. Additionally, the extent to which similar effects are found in countries without a democratic tradition is worth studying, since it is plausible that the present circumstances of health and economic adversity may offer them new opportunities for democratization.